Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

How useful is thrombocytosis in predicting an underlying cancer in primary care? a systematic review

How useful is thrombocytosis in predicting an underlying cancer in primary care? a systematic review Abstract Background. Although the association between raised platelet count (thrombocytosis) and cancer has been reported in primary and secondary care studies, UK GPs are unaware of it, and it is insufficiently evidenced for laboratories to identify and warn of it. This systematic review aimed to identify and collate evidence from studies that have investigated thrombocytosis as an early marker of cancer in primary care. Methods. EMBASE (OvidSP), Medline (Ovid), Web of Science and The Cochrane Library were searched for relevant studies. Eligible studies had reported estimates of the association between thrombocytosis and cancer, in adults aged ≥40 in a primary care setting. Raw data from included studies were used to calculate positive predictive values and likelihood ratios (LRs) for cancer. Results. Nine case–control studies were identified. Study quality was judged to be high. Included studies reported on the following cancer sites: colorectal, lung, ovary, bladder, kidney, pancreas, oesophago-gastric, uterus and breast. LRs indicated that thrombocytosis was a predictor of cancer in all sites except breast. In a consulting population, thrombocytosis is most highly predictive of lung and colorectal cancer. Conclusions. These results suggest that patients with thrombocytosis in primary care have an increased risk of cancer, and that some, but not all, cancers have raised platelets as an early marker. This finding is expected to be of use in primary care, for GPs receiving blood test results unexpectedly showing high platelet counts. Further research is needed to identify the cancers that are most strongly associated with thrombocytosis. Cancer, cancer diagnosis, family medicine, platelets, primary care, thrombocytosis. Introduction Early diagnosis of cancer is a leading objective, particularly in countries like the UK and Denmark with poor cancer outcomes compared to other European countries (1). It has been estimated that at least 5000 cancer deaths annually could be prevented in England by improvements in early diagnosis (2). Most cancers (~90% in the UK) present with symptoms to primary care (3). Many initiatives have been implemented to expedite diagnosis, with most targeting better investigative services, after the GP has considered the possibility of cancer. In the UK, these include 2-week wait clinics, open access GP investigation and NICE guidance (4). Much less attention has focussed on an earlier stage in primary care: identifying the very early risk markers of cancer to reduce diagnostic delay. A range of early symptoms and markers of cancer have been identified, and previous research in this area has resulted in the widespread use of tables of risk profiles for specific cancers in UK general practice (5–11). One factor which has received little attention is a raised platelet count. A raised platelet count, or thrombocytosis, has recently been identified as a marker of cancer, present before diagnosis. Previous studies have reported on the usefulness of platelet count as a prognostic tool in secondary care (12–24) and mechanisms that could underlie this association have been explored (25,26), but there is little evidence on the association in primary care where the initial suspicion of cancer is generally first made. This systematic review aimed to identify studies that have investigated whether adults aged ≥40 presenting with thrombocytosis in primary care are at greater risk of cancer than those with normal platelet counts and bring the results together in a narrative synthesis, to answer the general question of whether thrombocytosis is a marker of cancer. Methods Search strategy The search strategy was developed using the predefined terms ‘thrombocytosis’ or ‘platelet’ or ‘thrombocyte’ or ‘thrombocyte count’; and ‘cancer’ or ‘malignan*’; and ‘primary care’ or ‘primary medical care’ or ‘family practice’ or ‘family medicine’. The search was limited to English language articles published in the last 30 years, was first carried out in June 2015, and updated in February 2016. The following databases were searched: EMBASE (OvidSP), Medline (Ovid), Web of Science and The Cochrane Library. Forwards and backwards citation searching was carried out on included articles. Relevant experts in the field were asked if they had authored any relevant studies that had collected platelet count data. All results were exported to Endnote X5 and deduplicated. The study protocol can be accessed at http://hdl.handle.net/10871/20964. Study selection The search aimed to identify any study that had investigated the association between thrombocytosis and new diagnosis of cancer of any type in primary care. Inclusion criteria were as follows: adults aged ≥40, primary care setting and observational, case–control or cohort study or any literature review. Those under 40 years of age were excluded as cancers in this age group tend to be familial or atypical (indeed, no studies of lower age groups were identified in the searches). Studies that had investigated platelet count as a prognostic tool or guide for cancer therapy were excluded. Titles and abstracts were screened by two authors (SB and ES). Full text articles were retrieved and assessed for inclusion. Data analysis Study quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool (27). Studies were assessed for bias in four areas: patient selection, index test, reference standard and flow and timing. Raw data were extracted into custom-made data extraction forms; this was duplicated by SB and WH. Data were bought together in a narrative synthesis following the general framework set out in the Centre for Reviews and Dissemination’s Guidance for Undertaking Reviews in Health Care (28). Raw data from each study concerning the number of cases and healthy controls with thrombocytosis were used to calculate the likelihood ratio (LR: the probability of raised platelets in patients with cancer divided by the probability of raised platelets in healthy patients) and the positive predictive value (PPV) using Bayes’ Theorem. This approach to calculate PPVs is used with case–control study data as it accommodates prevalence in the general population, a necessary adjustment as prevalence is artificially high in case–control studies. Thus, PPVs were calculated as the LR multiplied by the prior odds of the disease (29); in this case, primary care incidence data were used to calculate prior odds. Subgroups were defined based on whether the cancer affects both sexes or one (males and females or females only), the data sources used by the studies and the source of the studies themselves (identified through literature search or contact with experts). Results The literature search identified 98 papers. After deduplicating, 79 titles and abstracts were screened for inclusion. Five full text articles were retrieved. The main reasons for exclusion were studies not being primary care-based, or not having a focus on cancer (Fig. 1). The two screening authors independently agreed on included studies. Figure 1. Open in new tabDownload slide Study selection process including reasons for exclusion of papers. Figure 1. Open in new tabDownload slide Study selection process including reasons for exclusion of papers. After reviewing full texts, four of the five articles met the inclusion criteria (5–8). All were case–control studies. Data from a further five case–control studies identified through contact with experts were added (10,30–33); these studies were not found by the literature search as they did not report platelet data, so did not contain any keywords related to thrombocytosis or platelets. However, following discussion with the authors of these studies, it became apparent that they had collected platelet data, and were able and willing to contribute raw data for the systematic review. These additional studies were carried out in the same research group as the four case–control studies identified by the literature search. In total, therefore, nine matched case–control studies were included. Study quality The overall quality of studies was high; the results of the QUADAS-2 assessment are shown in Table 1. Table 1. Quality assessment of each included study using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool (27) Reference . Risk of bias . Applicability concerns . Patient selection . Index test . Reference standard . Flow and timing . Patient selection . Index test . Reference standard . Walker et al. (33) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (6) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (7) ✓ ✓ ? ? ✓ ✓ ✓ Walker et al. (8) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (32) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (10) ✓ ✓ ? ? ✓ ✓ ✓ Hamilton et al. (30) ✓ ✓ ✓ ? ✓ ✓ ✓ Hamilton et al. (5) ✓ ✓ ✓ × ✓ ✓ ✓ Hamilton et al. (31) ✓ ✓ ✓ ? ✓ ✓ ✓ Reference . Risk of bias . Applicability concerns . Patient selection . Index test . Reference standard . Flow and timing . Patient selection . Index test . Reference standard . Walker et al. (33) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (6) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (7) ✓ ✓ ? ? ✓ ✓ ✓ Walker et al. (8) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (32) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (10) ✓ ✓ ? ? ✓ ✓ ✓ Hamilton et al. (30) ✓ ✓ ✓ ? ✓ ✓ ✓ Hamilton et al. (5) ✓ ✓ ✓ × ✓ ✓ ✓ Hamilton et al. (31) ✓ ✓ ✓ ? ✓ ✓ ✓ ✓ indicates a low risk of bias; ? indicates that it is unclear if there is bias and × indicates a high risk of bias. Open in new tab Table 1. Quality assessment of each included study using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool (27) Reference . Risk of bias . Applicability concerns . Patient selection . Index test . Reference standard . Flow and timing . Patient selection . Index test . Reference standard . Walker et al. (33) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (6) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (7) ✓ ✓ ? ? ✓ ✓ ✓ Walker et al. (8) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (32) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (10) ✓ ✓ ? ? ✓ ✓ ✓ Hamilton et al. (30) ✓ ✓ ✓ ? ✓ ✓ ✓ Hamilton et al. (5) ✓ ✓ ✓ × ✓ ✓ ✓ Hamilton et al. (31) ✓ ✓ ✓ ? ✓ ✓ ✓ Reference . Risk of bias . Applicability concerns . Patient selection . Index test . Reference standard . Flow and timing . Patient selection . Index test . Reference standard . Walker et al. (33) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (6) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (7) ✓ ✓ ? ? ✓ ✓ ✓ Walker et al. (8) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (32) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (10) ✓ ✓ ? ? ✓ ✓ ✓ Hamilton et al. (30) ✓ ✓ ✓ ? ✓ ✓ ✓ Hamilton et al. (5) ✓ ✓ ✓ × ✓ ✓ ✓ Hamilton et al. (31) ✓ ✓ ✓ ? ✓ ✓ ✓ ✓ indicates a low risk of bias; ? indicates that it is unclear if there is bias and × indicates a high risk of bias. Open in new tab The risk of bias was low across most items of the QUADAS-2 tool for three studies. One study showed a high risk of bias in the ‘flow and timing’ domain; in this study, only a small subsample of patients had platelet count values available for analysis. Most studies showed a small chance of bias in the ‘reference standard’ category, as these studies relied on cancer diagnoses recorded in patients’ electronic records, and did not check the accuracy of these diagnoses. The remaining studies had an unclear risk of bias in ‘flow and timing’ as it was not clear whether all patients had platelet counts available. Study characteristics The included studies are summarized in Table 2. Six of the nine case–control studies used data from the Clinical Practice Research Datalink (CPRD). This database holds electronic primary care records of patients from over 800 practices across the UK (34). Cases were all patients in the CPRD database with oesophago-gastric, pancreas, bladder, breast, kidney or uterine cancer diagnosed from 2000 to the year of study (by the identification of a code in the records: no histology was available), and up to five age, sex and practice-matched controls per case. The remaining three studies had included all cases of ovarian, colorectal or lung cancer within particular geographical areas, identified from Cancer Registry data (including histology) and from searching general practice records. Controls were age and sex matched patients randomly selected from the same practices. Table 2. Characteristics of included studies, including study type, cancer site, source of data used in the study, the number of cases and controls in each study and the proportion of females Reference . Cancer site . Study type . Data source . Cases . Controls . Walker et al. (33) Breast Case–control All women in the CPRD database diagnosed with breast cancer from 2000 to 2009 n = 4407 n = 21755 100% female 100% female Shephard et al. (6) Kidney Case–control All patients in the CPRD database diagnosed with kidney cancer from 2000 to 2009 n = 3183 n = 15707 38.7% female 40.2% female Stapley et al. (7) Oesophago-gastric Case–control All patients in the CPRD database diagnosed with oesophago-gastric cancer from 2000 to 2009 n = 7657 n = 37699 34.6% female 34.6% female Walker et al. (8) Uterine Case–control All women in the CPRD database diagnosed with uterine cancer from 2000 to 2009 n = 3166 n = 9537 100% female 100% female Shephard et al. (32) Bladder Case–control All patients in the CPRD database diagnosed with bladder cancer from 2000 to 2009 n = 4935 n = 24098 27.5% female 28.9% female Stapley et al. (10) Pancreatic Case–control All patients in the CPRD database diagnosed with pancreatic cancer from 2000 to 2009 n = 3635 n = 16459 52.0% female 52.0% female Hamilton et al. (30) Ovarian Case–control All women in Exeter and mid- and east-Devon diagnosed with ovarian cancer from 2000 to 2007, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 39 general practices in Exeter, mid-Devon and east-Devon, UK n = 212 n = 1060 100% female 100% female Hamilton et al. (31) Colorectal Case–control All patients in Exeter, Devon, diagnosed with colorectal cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 349 n = 1744 49.3% female 49.3% female Hamilton et al. (5) Lung Case–control All patients in Exeter, Devon, diagnosed with lung cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 247 n = 1235 31.9% female 31.9% female Reference . Cancer site . Study type . Data source . Cases . Controls . Walker et al. (33) Breast Case–control All women in the CPRD database diagnosed with breast cancer from 2000 to 2009 n = 4407 n = 21755 100% female 100% female Shephard et al. (6) Kidney Case–control All patients in the CPRD database diagnosed with kidney cancer from 2000 to 2009 n = 3183 n = 15707 38.7% female 40.2% female Stapley et al. (7) Oesophago-gastric Case–control All patients in the CPRD database diagnosed with oesophago-gastric cancer from 2000 to 2009 n = 7657 n = 37699 34.6% female 34.6% female Walker et al. (8) Uterine Case–control All women in the CPRD database diagnosed with uterine cancer from 2000 to 2009 n = 3166 n = 9537 100% female 100% female Shephard et al. (32) Bladder Case–control All patients in the CPRD database diagnosed with bladder cancer from 2000 to 2009 n = 4935 n = 24098 27.5% female 28.9% female Stapley et al. (10) Pancreatic Case–control All patients in the CPRD database diagnosed with pancreatic cancer from 2000 to 2009 n = 3635 n = 16459 52.0% female 52.0% female Hamilton et al. (30) Ovarian Case–control All women in Exeter and mid- and east-Devon diagnosed with ovarian cancer from 2000 to 2007, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 39 general practices in Exeter, mid-Devon and east-Devon, UK n = 212 n = 1060 100% female 100% female Hamilton et al. (31) Colorectal Case–control All patients in Exeter, Devon, diagnosed with colorectal cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 349 n = 1744 49.3% female 49.3% female Hamilton et al. (5) Lung Case–control All patients in Exeter, Devon, diagnosed with lung cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 247 n = 1235 31.9% female 31.9% female CPRD, Clinical Practice Research Datalink. Open in new tab Table 2. Characteristics of included studies, including study type, cancer site, source of data used in the study, the number of cases and controls in each study and the proportion of females Reference . Cancer site . Study type . Data source . Cases . Controls . Walker et al. (33) Breast Case–control All women in the CPRD database diagnosed with breast cancer from 2000 to 2009 n = 4407 n = 21755 100% female 100% female Shephard et al. (6) Kidney Case–control All patients in the CPRD database diagnosed with kidney cancer from 2000 to 2009 n = 3183 n = 15707 38.7% female 40.2% female Stapley et al. (7) Oesophago-gastric Case–control All patients in the CPRD database diagnosed with oesophago-gastric cancer from 2000 to 2009 n = 7657 n = 37699 34.6% female 34.6% female Walker et al. (8) Uterine Case–control All women in the CPRD database diagnosed with uterine cancer from 2000 to 2009 n = 3166 n = 9537 100% female 100% female Shephard et al. (32) Bladder Case–control All patients in the CPRD database diagnosed with bladder cancer from 2000 to 2009 n = 4935 n = 24098 27.5% female 28.9% female Stapley et al. (10) Pancreatic Case–control All patients in the CPRD database diagnosed with pancreatic cancer from 2000 to 2009 n = 3635 n = 16459 52.0% female 52.0% female Hamilton et al. (30) Ovarian Case–control All women in Exeter and mid- and east-Devon diagnosed with ovarian cancer from 2000 to 2007, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 39 general practices in Exeter, mid-Devon and east-Devon, UK n = 212 n = 1060 100% female 100% female Hamilton et al. (31) Colorectal Case–control All patients in Exeter, Devon, diagnosed with colorectal cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 349 n = 1744 49.3% female 49.3% female Hamilton et al. (5) Lung Case–control All patients in Exeter, Devon, diagnosed with lung cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 247 n = 1235 31.9% female 31.9% female Reference . Cancer site . Study type . Data source . Cases . Controls . Walker et al. (33) Breast Case–control All women in the CPRD database diagnosed with breast cancer from 2000 to 2009 n = 4407 n = 21755 100% female 100% female Shephard et al. (6) Kidney Case–control All patients in the CPRD database diagnosed with kidney cancer from 2000 to 2009 n = 3183 n = 15707 38.7% female 40.2% female Stapley et al. (7) Oesophago-gastric Case–control All patients in the CPRD database diagnosed with oesophago-gastric cancer from 2000 to 2009 n = 7657 n = 37699 34.6% female 34.6% female Walker et al. (8) Uterine Case–control All women in the CPRD database diagnosed with uterine cancer from 2000 to 2009 n = 3166 n = 9537 100% female 100% female Shephard et al. (32) Bladder Case–control All patients in the CPRD database diagnosed with bladder cancer from 2000 to 2009 n = 4935 n = 24098 27.5% female 28.9% female Stapley et al. (10) Pancreatic Case–control All patients in the CPRD database diagnosed with pancreatic cancer from 2000 to 2009 n = 3635 n = 16459 52.0% female 52.0% female Hamilton et al. (30) Ovarian Case–control All women in Exeter and mid- and east-Devon diagnosed with ovarian cancer from 2000 to 2007, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 39 general practices in Exeter, mid-Devon and east-Devon, UK n = 212 n = 1060 100% female 100% female Hamilton et al. (31) Colorectal Case–control All patients in Exeter, Devon, diagnosed with colorectal cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 349 n = 1744 49.3% female 49.3% female Hamilton et al. (5) Lung Case–control All patients in Exeter, Devon, diagnosed with lung cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 247 n = 1235 31.9% female 31.9% female CPRD, Clinical Practice Research Datalink. Open in new tab All studies excluded patients with metastatic cancer from an earlier primary site (although included patients with metastatic disease at the time of diagnosis), patients for whom no matching controls were available, patients with no data available in the year prior to diagnosis, duplicates or controls whose matching case had been excluded. Uterine cancer controls with a hysterectomy prior to the cancer diagnosis date of their associated cases (8) and patients with a mastectomy >3 months prior to their diagnosis date (33) were excluded. All studies defined thrombocytosis as a platelet count above 400 or 450×109/l depending on the local laboratory definition. All studies had included adults aged ≥40 years. Association between thrombocytosis and cancer Four studies had carried out multivariable analyses and found a statistically significant association (P < 0.05) between thrombocytosis and their specific cancer site (5–8). The adjusted odds ratios (ORs) as presented in the publications were as follows: for lung cancer, 9.3 [95% confidence interval (CI) 3.4–26]; for kidney cancer, 2.2 (95% CI 1.7–2.7); for oesophago-gastric cancer 2.4 (95% CI 2.0–2.9) and for uterine cancer 1.50 (95% CI 1.00–2.25). Four of the remaining five studies reported no significant association between thrombocytosis and cancer (10, 30, 31, 33); these were the studies identified through contact with experts as this non-significant association was not reported in the publication The fifth study collected but did not analyse platelet count data (32). The independently calculated LRs for each cancer site are presented in Figure 2; all with the exception of breast cancer indicate an increased probability of malignancy in patients with a blood test showing thrombocytosis. Figure 2. Open in new tabDownload slide The number of cases and controls included in each study, and the number (n) of these with thrombocytosis in the year prior to diagnosis. Independently calculated likelihood ratios, with 95% confidence intervals, are estimated, and these are shown graphically in a forest plot. Figure 2. Open in new tabDownload slide The number of cases and controls included in each study, and the number (n) of these with thrombocytosis in the year prior to diagnosis. Independently calculated likelihood ratios, with 95% confidence intervals, are estimated, and these are shown graphically in a forest plot. The PPVs for each cancer site are shown in Table 3. LRs were highest in ovarian, lung, kidney, colorectal and oesophago-gastric cancer, but due to differences in the prevalence of these cancers in the consulting population, lung and colorectal (the most common cancers of the nine) had the highest PPVs for cancer. Ovarian cancer, with the greatest LR of 14.61, had a relatively small PPV of 0.65 which reflects the relative rarity of this diagnosis. Table 3. Analysis of data from included studies. The table details the number of cases and controls in each study and how many of these had a blood test result showing thrombocytosis within the year prior to diagnosis. Independently calculated positive predictive values (PPVs) are presented along with 95% confidence intervals (CI). Reference . Study source . Cancer site . Cases n, n (%) with thrombocytosis . Controls n, n (%) with thrombocytosis . Positive predictive values % (95% CI) . Hamilton et al. (5) Literature search Lung 247 1235 1.63 34 (13.8) 19 (1.5) (0.92–2.90) Shephard et al. (6) Literature search Kidney 3183 15707 0.17 348 (10.9) 251 (1.6) (0.15–0.20) Stapley et al. (7) Literature search Oesophago-gastric 7657 37699 0.47 707 (9.2) 568 (1.5) (0.42–0.52) Walker et al. (8) Literature search Uterine 3166 9537 0.08 110 (3.5) 207 (2.2) (0.07–0.11) Walker et al. (33) Contact with experts Breast 4407 21755 0.38 91 (2.1) 369 (1.7) (0.31–0.48) Shephard et al. (32) Contact with experts Bladder 4935 24098 0.10 156 (3.2) 247 (1.0) (0.08–0.12) Stapley et al. (10) Contact with experts Pancreatic 3635 16459 0.13 214 (5.9) 222 (1.3) (0.11–0.15) Hamilton et al. (30) Contact with experts Ovarian 212 1060 0.65 26 (12.3) 9 (0.8) (0.31–1.36) Stapley et al. (11) Contact with experts Colorectal 349 1744 1.39 48 (13.8) 42 (2.4) (0.94–2.09) Reference . Study source . Cancer site . Cases n, n (%) with thrombocytosis . Controls n, n (%) with thrombocytosis . Positive predictive values % (95% CI) . Hamilton et al. (5) Literature search Lung 247 1235 1.63 34 (13.8) 19 (1.5) (0.92–2.90) Shephard et al. (6) Literature search Kidney 3183 15707 0.17 348 (10.9) 251 (1.6) (0.15–0.20) Stapley et al. (7) Literature search Oesophago-gastric 7657 37699 0.47 707 (9.2) 568 (1.5) (0.42–0.52) Walker et al. (8) Literature search Uterine 3166 9537 0.08 110 (3.5) 207 (2.2) (0.07–0.11) Walker et al. (33) Contact with experts Breast 4407 21755 0.38 91 (2.1) 369 (1.7) (0.31–0.48) Shephard et al. (32) Contact with experts Bladder 4935 24098 0.10 156 (3.2) 247 (1.0) (0.08–0.12) Stapley et al. (10) Contact with experts Pancreatic 3635 16459 0.13 214 (5.9) 222 (1.3) (0.11–0.15) Hamilton et al. (30) Contact with experts Ovarian 212 1060 0.65 26 (12.3) 9 (0.8) (0.31–1.36) Stapley et al. (11) Contact with experts Colorectal 349 1744 1.39 48 (13.8) 42 (2.4) (0.94–2.09) Open in new tab Table 3. Analysis of data from included studies. The table details the number of cases and controls in each study and how many of these had a blood test result showing thrombocytosis within the year prior to diagnosis. Independently calculated positive predictive values (PPVs) are presented along with 95% confidence intervals (CI). Reference . Study source . Cancer site . Cases n, n (%) with thrombocytosis . Controls n, n (%) with thrombocytosis . Positive predictive values % (95% CI) . Hamilton et al. (5) Literature search Lung 247 1235 1.63 34 (13.8) 19 (1.5) (0.92–2.90) Shephard et al. (6) Literature search Kidney 3183 15707 0.17 348 (10.9) 251 (1.6) (0.15–0.20) Stapley et al. (7) Literature search Oesophago-gastric 7657 37699 0.47 707 (9.2) 568 (1.5) (0.42–0.52) Walker et al. (8) Literature search Uterine 3166 9537 0.08 110 (3.5) 207 (2.2) (0.07–0.11) Walker et al. (33) Contact with experts Breast 4407 21755 0.38 91 (2.1) 369 (1.7) (0.31–0.48) Shephard et al. (32) Contact with experts Bladder 4935 24098 0.10 156 (3.2) 247 (1.0) (0.08–0.12) Stapley et al. (10) Contact with experts Pancreatic 3635 16459 0.13 214 (5.9) 222 (1.3) (0.11–0.15) Hamilton et al. (30) Contact with experts Ovarian 212 1060 0.65 26 (12.3) 9 (0.8) (0.31–1.36) Stapley et al. (11) Contact with experts Colorectal 349 1744 1.39 48 (13.8) 42 (2.4) (0.94–2.09) Reference . Study source . Cancer site . Cases n, n (%) with thrombocytosis . Controls n, n (%) with thrombocytosis . Positive predictive values % (95% CI) . Hamilton et al. (5) Literature search Lung 247 1235 1.63 34 (13.8) 19 (1.5) (0.92–2.90) Shephard et al. (6) Literature search Kidney 3183 15707 0.17 348 (10.9) 251 (1.6) (0.15–0.20) Stapley et al. (7) Literature search Oesophago-gastric 7657 37699 0.47 707 (9.2) 568 (1.5) (0.42–0.52) Walker et al. (8) Literature search Uterine 3166 9537 0.08 110 (3.5) 207 (2.2) (0.07–0.11) Walker et al. (33) Contact with experts Breast 4407 21755 0.38 91 (2.1) 369 (1.7) (0.31–0.48) Shephard et al. (32) Contact with experts Bladder 4935 24098 0.10 156 (3.2) 247 (1.0) (0.08–0.12) Stapley et al. (10) Contact with experts Pancreatic 3635 16459 0.13 214 (5.9) 222 (1.3) (0.11–0.15) Hamilton et al. (30) Contact with experts Ovarian 212 1060 0.65 26 (12.3) 9 (0.8) (0.31–1.36) Stapley et al. (11) Contact with experts Colorectal 349 1744 1.39 48 (13.8) 42 (2.4) (0.94–2.09) Open in new tab Generally, studies that used CPRD data achieved larger sample sizes than those that used general practice records. The former included all patients diagnosed with the cancer of interest from any UK practice whereas the latter included patients diagnosed with the cancer of interest from set geographical areas within South-West England. Discussion Summary This is the first systematic review to identify and collate results from studies investigating the association between thrombocytosis and diagnosis of cancer in primary care. These results suggest that patients with thrombocytosis in primary care have an increased risk of cancer, and that some types of cancer are more strongly associated with thrombocytosis than others. All but one of the nine cancer sites had significant LRs calculated from raw data; for four of the nine studies, however, significance was not retained in multivariable models. Although this does not detract from the overall conclusion of an association between cancer and thrombocytosis, it does suggest that the results of this review should be interpreted with caution and supports the view that the association only exists for certain types of cancer. There did not appear to be any biological link between those cancer sites that did and did not have a significant association with thrombocytosis in multivariable models. When considering the anatomy of cancer sites, there were counterintuitive results; firstly, although colorectal and oesophago-gastric cancer, both of the digestive system, had similar ORs for cancer (6.5 and 5.9, respectively), only oesophago-gastric cancer retained significance in published multivariable models. In our independent analysis, both sites had LRs of ~5, but the PPV for colorectal cancer (1.39%) was higher than that for oesophago-gastric cancer (0.47%); this difference can be explained by the fact that colorectal cancer is more commonly diagnosed than oesophago-gastric. Secondly, uterine and ovarian cancer had somewhat disparate results. Uterine cancer alone retained significance both in the multivariable model, and in our independent analysis (LR 14.61). Key differences in the way these types of cancer develop or manifest themselves may underlie these observations. Further investigation of this could provide evidence for the biological mechanisms that underlie the association between thrombocytosis and cancer. Strengths and limitations The quality of studies included in this review was judged to be high, and the majority used CPRD data. A long-established high-quality data source, the CPRD holds primary care records for patients from 684 GP practices, covering 8.8% of the UK population with an even geographical distribution. This largely representative sample can yield studies with widely applicable results. It is particularly relevant that the study data are taken from primary care as it is in this environment that the study results should be clinically useful. The validity of CPRD data has been found to be high in two recent systematic reviews (35,36), although none of the studies included in these reviews addressed cancer data specifically. The CPRD studies include all patients in the UK within CPRD-registered practices who were diagnosed with the cancer of choice rather than patients within a restricted geographical area, as in the non-CPRD studies. This reduces the element of selection bias which is inherent in many case–control studies. It also enables much larger studies, and consequently more accurate and reliable results. However, CRPD cases are based on electronic records of cancer diagnoses alone; whereas in the non-CPRD studies, electronic records from the Cancer Registry were used as well as articles general practice records which included histology reports and were therefore more likely to be accurate. In terms of the exposure variable, the platelet counts recorded in the CPRD are electronic and automatically transmitted from laboratory to the patient records, reducing the chance of human error and subsequent risk of classification bias. Where platelet count data were missing for patients in CPRD studies, patients were assumed to not have thrombocytosis. In contrast, the non-CPRD studies relied on manual checking and recording of platelet counts, more open to sources of bias and error. Overall, the CPRD studies can be considered higher quality than the non-CPRD studies; these show a positive association between thrombocytosis and cancer, although there is marked variation between different cancer sites. Although these results can readily be applied to UK general practice, the lack of any non-UK studies limits the extent to which these findings can be generalized outside the UK. There is also a chance that relevant data may not have been identified due to publication bias. The majority of studies that contributed raw data to this study were not identified by the literature review because they did not report non-significant findings. Although the lead author of this systematic review contacted relevant experts who have carried out similar studies to ask if they had collected platelet count data (they had not), we were only able to contact those we were aware of through networking, conference attendance and general knowledge of the field. It is possible that other researchers unknown to this team are carrying out similar work and have collected data, but that this has remained unpublished due to a lack of significant findings. Another potential source of bias lies in the fact that WH, an author of this article, is an author in all nine studies included in this systematic review, and ES of three. However, the search strategy and quality assessment were carried out independently by SB and OU, who were not involved in any of the nine included studies. Comparison with existing literature The biological processes behind the thrombocytosis–cancer association have been studied in patients post-diagnosis (25,26) but the effect is not fully understood. It is also uncertain whether the mechanisms behind the association in secondary care would apply in primary care, before diagnosis is made. The association between uterine cancer and thrombocytosis is supported by a study from Stone et al. (2012) which found that 31% of patients with ovarian cancer had thrombocytosis at the point of diagnosis (38). Implications for research and practice This systematic review suggests that thrombocytosis is an early marker of some cancers in primary care and will raise awareness of this marker among GPs. This finding can be of use in primary care for GPs receiving blood results unexpectedly showing high platelet counts. Simply adding the PPVs from Table 3 suggests that the PPV from thrombocytosis is at least 5% for any of the nine cancers (several cancer sites have not been reported and any association with thrombocytosis with them would increase this figure). It would be useful to study thrombocytosis across cancer as a whole to give a more accurate estimate of the risk, but it is very likely to remain an important feature, especially now the UK has moved to rapid investigation of any cancer risk exceeding 3% (37). Any further study should also determine which specific cancers present with raised platelets as an early marker allowing targeted investigation. Crucially, this review does not suggest that platelet count should be specifically used as a diagnostic test for cancer, or as a screening tool. The current body of evidence only supports clinicians considering cancer as a diagnosis if a blood test result shows thrombocytosis. Future studies should compare risk for specific patient subgroups, including men and women, those in different age groups, and at various levels of elevated platelets. Ideally, studies could suggest the optimal platelet count to distinguish between those who do and do not have cancer. It may also be possible to combine thrombocytosis with other early markers of cancer to develop cancer-specific risk values for combinations of symptoms. Declaration Funding: the Policy Research Unit in Cancer Awareness, Screening and Early Diagnosis receives funding for a research programme from the Department of Health Policy Research Programme. It is collaboration between researchers from seven institutions (Queen Mary University of London, UCL, King’s College London, London School of Hygiene and Tropical Medicine, Hull York Medical School, Durham University and University of Exeter). OU is funded, and WH is part-funded by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care (CLAHRC) for the South West Peninsula at the Royal Devon and Exeter NHS Foundation Trust. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health in England. Ethical approval: none. Conflict of interest: none. Acknowledgements Many thanks to Sal Stapley and Sarah Walker for sharing data from their studies. Thanks also to Tim Starkey for producing the forest plot in Matlab. References 1. Coleman MP Forman D Bryant H et al. Cancer survival in Australia, Canada, Denmark, Norway, Sweden, and the UK, 1995–2007 (the International Cancer Benchmarking Partnership): an analysis of population-based cancer registry data . Lancet 2011 ; 377 : 127 – 38 . Google Scholar Crossref Search ADS PubMed WorldCat 2. Richards MA . The size of the prize for earlier diagnosis of cancer in England . Br J Cancer 2009 ; 101 : S125 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat 3. Hamilton W . Cancer diagnosis in primary care . Br J Gen Pract 2010 ; 60 : 121 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 4. NICE . Referral Guidelines for Suspected Cancer . London, UK : NICE , 2005 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 5. Hamilton W Peters TJ Round A Sharp D . What are the clinical features of lung cancer before the diagnosis is made? A population based case-control study . Thorax 2005 ; 60 : 1059 – 65 . Google Scholar Crossref Search ADS PubMed WorldCat 6. Shephard E Neal R Rose P et al. Clinical features of kidney cancer in primary care: a case-control study using primary care records . Br J Gen Pract 2013 ; 63 : e250 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat 7. Stapley S Peters TJ Neal RD et al. The risk of oesophago-gastric cancer in symptomatic patients in primary care: a large case-control study using electronic records . Br J Cancer 2013 ; 108 : 25 – 31 . Google Scholar Crossref Search ADS PubMed WorldCat 8. Walker S Hyde C Hamilton W . Risk of uterine cancer in symptomatic women in primary care: case-control study using electronic records . Br J Gen Pract 2013 ; 63 : e643 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 9. Barrett J Sharp DJ Stapley S et al. Pathways to the diagnosis of ovarian cancer in the UK: a cohort study in primary care . BJOG 2010 ; 117 : 610 – 4 . Google Scholar Crossref Search ADS PubMed WorldCat 10. Stapley S Peters TJ Neal RD et al. The risk of pancreatic cancer in symptomatic patients in primary care: a large case-control study using electronic records . Br J Cancer 2012 ; 106 : 1940 – 4 . Google Scholar Crossref Search ADS PubMed WorldCat 11. Stapley S Peters TJ Sharp D Hamilton W . The mortality of colorectal cancer in relation to the initial symptom at presentation to primary care and to the duration of symptoms: a cohort study using medical records . Br J Cancer 2006 ; 95 : 1321 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat 12. Aoe K Hiraki A Ueoka H et al. Thrombocytosis as a useful prognostic indicator in patients with lung cancer . Respiration 2004 ; 71 : 170 – 3 . Google Scholar Crossref Search ADS PubMed WorldCat 13. Bensalah K Leray E Fergelot P et al. Prognostic value of thrombocytosis in renal cell carcinoma . J Urol 2006 ; 175 : 859 – 63 . Google Scholar Crossref Search ADS PubMed WorldCat 14. Carr BI Guerra V . Thrombocytosis and hepatocellular carcinoma . Dig Dis Sci 2013 ; 58 : 1790 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat 15. Cravioto-Villanueva A Luna-Perez P Gutierrez-de la Barrera M et al. Thrombocytosis as a predictor of distant recurrence in patients with rectal cancer . Arch Med Res 2012 ; 43 : 305 – 11 . Google Scholar Crossref Search ADS PubMed WorldCat 16. Gogus C Baltaci S Filiz E et al. Significance of thrombocytosis for determining prognosis in patients with localized renal cell carcinoma . Urology 2004 ; 63 : 447 – 50 . Google Scholar Crossref Search ADS PubMed WorldCat 17. Gucer F Tamussino K Keil F et al. Thrombocytosis in gynecologic malignancies . Anticancer Res 2004 ; 24 : 2053 – 9 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 18. Holgersson G Sandelin M Hoye E et al. Swedish lung cancer radiation study group: the prognostic value of anaemia, thrombocytosis and leukocytosis at time of diagnosis in patients with non-small cell lung cancer . Med Oncol 2012 ; 29 : 3176 – 82 . Google Scholar Crossref Search ADS PubMed WorldCat 19. Hwang SG Kim KM Cheong JH et al. . Impact of pretreatment thrombocytosis on blood-borne metastasis and prognosis of gastric cancer . Eur J Surg Oncol 2012 ; 38 : 562 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat 20. Inoue K Kohashikawa K Suzuki S et al. Prognostic significance of thrombocytosis in renal cell carcinoma patients . Int J Urol 2004 ; 11 : 364 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat 21. Ishizuka M Nagata H Takagi K et al. Preoperative thrombocytosis is associated with survival after surgery for colorectal cancer . J Surg Oncol 2012 ; 106 : 887 – 91 . Google Scholar Crossref Search ADS PubMed WorldCat 22. Li AJ Madden AC Cass I et al. The prognostic significance of thrombocytosis in epithelial ovarian carcinoma . Gynecol Oncol 2004 ; 92 : 211 – 4 . Google Scholar Crossref Search ADS PubMed WorldCat 23. Pedersen LM Milman N . Prognostic significance of thrombocytosis in patients with primary lung cancer . Eur Respir J 1996 ; 9 : 1826 – 30 . Google Scholar Crossref Search ADS PubMed WorldCat 24. Suppiah R Shaheen PE Elson P et al. Thrombocytosis as a prognostic factor for survival in patients with metastatic renal cell carcinoma . Cancer 2006 ; 107 : 1793 – 800 . Google Scholar Crossref Search ADS PubMed WorldCat 25. Arslan Ç Coşkun HŞ . Thrombocytosis in solid tumors: review of the literature . Turk J Haematol 2005 ; 22 : 59 – 64 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 26. Nash GF Turner LF Scully MF Kakkar AK . Platelets and cancer . Lancet Oncol 2002 ; 3 : 425 – 30 . Google Scholar Crossref Search ADS PubMed WorldCat 27. Whiting PF Rutjes AW Westwood ME et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies . Ann Intern Med 2011 ; 155 : 529 – 36 . Google Scholar Crossref Search ADS PubMed WorldCat 28. Centre for Reviews and Dissemination . Systematic Reviews: CRD’s Guidance for Undertaking Reviews in Health Care . York, UK : CRD, University of York , 2008 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 29. Knottnerus JA. The Evidence Base of Clinical Diagnosis . London, UK : BMJ Books , 2002 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 30. Hamilton W Peters TJ Bankhead C Sharp D . Risk of ovarian cancer in women with symptoms in primary care: population based case-control study . Br Med J 2009 ; 339 : b2998 . Google Scholar Crossref Search ADS WorldCat 31. Hamilton W Round A Sharp D Peters TJ . Clinical features of colorectal cancer before diagnosis: a population-based case-control study . Br J Cancer 2005 ; 93 : 399 – 405 . Google Scholar Crossref Search ADS PubMed WorldCat 32. Shephard E Stapley S Neal RD et al. Clinical features of bladder cancer in primary care . Br J Gen Pract 2012 ; 62 : e598 – 604 . Google Scholar Crossref Search ADS PubMed WorldCat 33. Walker S Hyde C Hamilton W . Risk of breast cancer in symptomatic women in primary care: a case control study using primary care records . Br J Gen Pract 2014 ; 64 : e788 – 93 . Google Scholar Crossref Search ADS PubMed WorldCat 34. Williams T van Staa T Puri S Eaton S . Recent advances in the utility and use of the general practice research database as an example of a UK primary care data resource . Ther Adv Drug Saf 2012 ; 3 : 89 – 99 . Google Scholar Crossref Search ADS PubMed WorldCat 35. Herrett E Thomas SL Schoonen WM et al. Validation and validity of diagnoses in the general practice research database: a systematic review . Br J Clin Pharmacol 2010 ; 69 : 4 – 14 . Google Scholar Crossref Search ADS PubMed WorldCat 36. Khan NF Harrison SE Rose PW . Validity of diagnostic coding within the general practice research database: a systematic review . Br J Gen Pract 2010 ; 60 : e128 – 36 . Google Scholar Crossref Search ADS PubMed WorldCat 37. NICE . Suspected Cancer: Recognition and Referral . London, UK : NICE , 2015 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 38. Stone R Nick A McNeish I et al. Paraneoplastic thrombocytosis in ovarian cancer . NEJM 2012 ; 366 : 610 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Author notes " *Correspondence to Sarah Bailey, Primary Care Diagnostics, University of Exeter Medical School, College House, St Luke’s Campus, University of Exeter, Exeter, Devon EX1 2LU, UK; E-mail: s.e.r.bailey@exeter.ac.uk © The Author 2016. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Family Practice Oxford University Press

How useful is thrombocytosis in predicting an underlying cancer in primary care? a systematic review

Loading next page...
 
/lp/oxford-university-press/how-useful-is-thrombocytosis-in-predicting-an-underlying-cancer-in-5TzLLGIiLg

References (46)

Publisher
Oxford University Press
Copyright
© The Author 2016. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0263-2136
eISSN
1460-2229
DOI
10.1093/fampra/cmw100
Publisher site
See Article on Publisher Site

Abstract

Abstract Background. Although the association between raised platelet count (thrombocytosis) and cancer has been reported in primary and secondary care studies, UK GPs are unaware of it, and it is insufficiently evidenced for laboratories to identify and warn of it. This systematic review aimed to identify and collate evidence from studies that have investigated thrombocytosis as an early marker of cancer in primary care. Methods. EMBASE (OvidSP), Medline (Ovid), Web of Science and The Cochrane Library were searched for relevant studies. Eligible studies had reported estimates of the association between thrombocytosis and cancer, in adults aged ≥40 in a primary care setting. Raw data from included studies were used to calculate positive predictive values and likelihood ratios (LRs) for cancer. Results. Nine case–control studies were identified. Study quality was judged to be high. Included studies reported on the following cancer sites: colorectal, lung, ovary, bladder, kidney, pancreas, oesophago-gastric, uterus and breast. LRs indicated that thrombocytosis was a predictor of cancer in all sites except breast. In a consulting population, thrombocytosis is most highly predictive of lung and colorectal cancer. Conclusions. These results suggest that patients with thrombocytosis in primary care have an increased risk of cancer, and that some, but not all, cancers have raised platelets as an early marker. This finding is expected to be of use in primary care, for GPs receiving blood test results unexpectedly showing high platelet counts. Further research is needed to identify the cancers that are most strongly associated with thrombocytosis. Cancer, cancer diagnosis, family medicine, platelets, primary care, thrombocytosis. Introduction Early diagnosis of cancer is a leading objective, particularly in countries like the UK and Denmark with poor cancer outcomes compared to other European countries (1). It has been estimated that at least 5000 cancer deaths annually could be prevented in England by improvements in early diagnosis (2). Most cancers (~90% in the UK) present with symptoms to primary care (3). Many initiatives have been implemented to expedite diagnosis, with most targeting better investigative services, after the GP has considered the possibility of cancer. In the UK, these include 2-week wait clinics, open access GP investigation and NICE guidance (4). Much less attention has focussed on an earlier stage in primary care: identifying the very early risk markers of cancer to reduce diagnostic delay. A range of early symptoms and markers of cancer have been identified, and previous research in this area has resulted in the widespread use of tables of risk profiles for specific cancers in UK general practice (5–11). One factor which has received little attention is a raised platelet count. A raised platelet count, or thrombocytosis, has recently been identified as a marker of cancer, present before diagnosis. Previous studies have reported on the usefulness of platelet count as a prognostic tool in secondary care (12–24) and mechanisms that could underlie this association have been explored (25,26), but there is little evidence on the association in primary care where the initial suspicion of cancer is generally first made. This systematic review aimed to identify studies that have investigated whether adults aged ≥40 presenting with thrombocytosis in primary care are at greater risk of cancer than those with normal platelet counts and bring the results together in a narrative synthesis, to answer the general question of whether thrombocytosis is a marker of cancer. Methods Search strategy The search strategy was developed using the predefined terms ‘thrombocytosis’ or ‘platelet’ or ‘thrombocyte’ or ‘thrombocyte count’; and ‘cancer’ or ‘malignan*’; and ‘primary care’ or ‘primary medical care’ or ‘family practice’ or ‘family medicine’. The search was limited to English language articles published in the last 30 years, was first carried out in June 2015, and updated in February 2016. The following databases were searched: EMBASE (OvidSP), Medline (Ovid), Web of Science and The Cochrane Library. Forwards and backwards citation searching was carried out on included articles. Relevant experts in the field were asked if they had authored any relevant studies that had collected platelet count data. All results were exported to Endnote X5 and deduplicated. The study protocol can be accessed at http://hdl.handle.net/10871/20964. Study selection The search aimed to identify any study that had investigated the association between thrombocytosis and new diagnosis of cancer of any type in primary care. Inclusion criteria were as follows: adults aged ≥40, primary care setting and observational, case–control or cohort study or any literature review. Those under 40 years of age were excluded as cancers in this age group tend to be familial or atypical (indeed, no studies of lower age groups were identified in the searches). Studies that had investigated platelet count as a prognostic tool or guide for cancer therapy were excluded. Titles and abstracts were screened by two authors (SB and ES). Full text articles were retrieved and assessed for inclusion. Data analysis Study quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool (27). Studies were assessed for bias in four areas: patient selection, index test, reference standard and flow and timing. Raw data were extracted into custom-made data extraction forms; this was duplicated by SB and WH. Data were bought together in a narrative synthesis following the general framework set out in the Centre for Reviews and Dissemination’s Guidance for Undertaking Reviews in Health Care (28). Raw data from each study concerning the number of cases and healthy controls with thrombocytosis were used to calculate the likelihood ratio (LR: the probability of raised platelets in patients with cancer divided by the probability of raised platelets in healthy patients) and the positive predictive value (PPV) using Bayes’ Theorem. This approach to calculate PPVs is used with case–control study data as it accommodates prevalence in the general population, a necessary adjustment as prevalence is artificially high in case–control studies. Thus, PPVs were calculated as the LR multiplied by the prior odds of the disease (29); in this case, primary care incidence data were used to calculate prior odds. Subgroups were defined based on whether the cancer affects both sexes or one (males and females or females only), the data sources used by the studies and the source of the studies themselves (identified through literature search or contact with experts). Results The literature search identified 98 papers. After deduplicating, 79 titles and abstracts were screened for inclusion. Five full text articles were retrieved. The main reasons for exclusion were studies not being primary care-based, or not having a focus on cancer (Fig. 1). The two screening authors independently agreed on included studies. Figure 1. Open in new tabDownload slide Study selection process including reasons for exclusion of papers. Figure 1. Open in new tabDownload slide Study selection process including reasons for exclusion of papers. After reviewing full texts, four of the five articles met the inclusion criteria (5–8). All were case–control studies. Data from a further five case–control studies identified through contact with experts were added (10,30–33); these studies were not found by the literature search as they did not report platelet data, so did not contain any keywords related to thrombocytosis or platelets. However, following discussion with the authors of these studies, it became apparent that they had collected platelet data, and were able and willing to contribute raw data for the systematic review. These additional studies were carried out in the same research group as the four case–control studies identified by the literature search. In total, therefore, nine matched case–control studies were included. Study quality The overall quality of studies was high; the results of the QUADAS-2 assessment are shown in Table 1. Table 1. Quality assessment of each included study using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool (27) Reference . Risk of bias . Applicability concerns . Patient selection . Index test . Reference standard . Flow and timing . Patient selection . Index test . Reference standard . Walker et al. (33) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (6) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (7) ✓ ✓ ? ? ✓ ✓ ✓ Walker et al. (8) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (32) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (10) ✓ ✓ ? ? ✓ ✓ ✓ Hamilton et al. (30) ✓ ✓ ✓ ? ✓ ✓ ✓ Hamilton et al. (5) ✓ ✓ ✓ × ✓ ✓ ✓ Hamilton et al. (31) ✓ ✓ ✓ ? ✓ ✓ ✓ Reference . Risk of bias . Applicability concerns . Patient selection . Index test . Reference standard . Flow and timing . Patient selection . Index test . Reference standard . Walker et al. (33) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (6) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (7) ✓ ✓ ? ? ✓ ✓ ✓ Walker et al. (8) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (32) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (10) ✓ ✓ ? ? ✓ ✓ ✓ Hamilton et al. (30) ✓ ✓ ✓ ? ✓ ✓ ✓ Hamilton et al. (5) ✓ ✓ ✓ × ✓ ✓ ✓ Hamilton et al. (31) ✓ ✓ ✓ ? ✓ ✓ ✓ ✓ indicates a low risk of bias; ? indicates that it is unclear if there is bias and × indicates a high risk of bias. Open in new tab Table 1. Quality assessment of each included study using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool (27) Reference . Risk of bias . Applicability concerns . Patient selection . Index test . Reference standard . Flow and timing . Patient selection . Index test . Reference standard . Walker et al. (33) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (6) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (7) ✓ ✓ ? ? ✓ ✓ ✓ Walker et al. (8) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (32) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (10) ✓ ✓ ? ? ✓ ✓ ✓ Hamilton et al. (30) ✓ ✓ ✓ ? ✓ ✓ ✓ Hamilton et al. (5) ✓ ✓ ✓ × ✓ ✓ ✓ Hamilton et al. (31) ✓ ✓ ✓ ? ✓ ✓ ✓ Reference . Risk of bias . Applicability concerns . Patient selection . Index test . Reference standard . Flow and timing . Patient selection . Index test . Reference standard . Walker et al. (33) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (6) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (7) ✓ ✓ ? ? ✓ ✓ ✓ Walker et al. (8) ✓ ✓ ? ? ✓ ✓ ✓ Shephard et al. (32) ✓ ✓ ? ? ✓ ✓ ✓ Stapley et al. (10) ✓ ✓ ? ? ✓ ✓ ✓ Hamilton et al. (30) ✓ ✓ ✓ ? ✓ ✓ ✓ Hamilton et al. (5) ✓ ✓ ✓ × ✓ ✓ ✓ Hamilton et al. (31) ✓ ✓ ✓ ? ✓ ✓ ✓ ✓ indicates a low risk of bias; ? indicates that it is unclear if there is bias and × indicates a high risk of bias. Open in new tab The risk of bias was low across most items of the QUADAS-2 tool for three studies. One study showed a high risk of bias in the ‘flow and timing’ domain; in this study, only a small subsample of patients had platelet count values available for analysis. Most studies showed a small chance of bias in the ‘reference standard’ category, as these studies relied on cancer diagnoses recorded in patients’ electronic records, and did not check the accuracy of these diagnoses. The remaining studies had an unclear risk of bias in ‘flow and timing’ as it was not clear whether all patients had platelet counts available. Study characteristics The included studies are summarized in Table 2. Six of the nine case–control studies used data from the Clinical Practice Research Datalink (CPRD). This database holds electronic primary care records of patients from over 800 practices across the UK (34). Cases were all patients in the CPRD database with oesophago-gastric, pancreas, bladder, breast, kidney or uterine cancer diagnosed from 2000 to the year of study (by the identification of a code in the records: no histology was available), and up to five age, sex and practice-matched controls per case. The remaining three studies had included all cases of ovarian, colorectal or lung cancer within particular geographical areas, identified from Cancer Registry data (including histology) and from searching general practice records. Controls were age and sex matched patients randomly selected from the same practices. Table 2. Characteristics of included studies, including study type, cancer site, source of data used in the study, the number of cases and controls in each study and the proportion of females Reference . Cancer site . Study type . Data source . Cases . Controls . Walker et al. (33) Breast Case–control All women in the CPRD database diagnosed with breast cancer from 2000 to 2009 n = 4407 n = 21755 100% female 100% female Shephard et al. (6) Kidney Case–control All patients in the CPRD database diagnosed with kidney cancer from 2000 to 2009 n = 3183 n = 15707 38.7% female 40.2% female Stapley et al. (7) Oesophago-gastric Case–control All patients in the CPRD database diagnosed with oesophago-gastric cancer from 2000 to 2009 n = 7657 n = 37699 34.6% female 34.6% female Walker et al. (8) Uterine Case–control All women in the CPRD database diagnosed with uterine cancer from 2000 to 2009 n = 3166 n = 9537 100% female 100% female Shephard et al. (32) Bladder Case–control All patients in the CPRD database diagnosed with bladder cancer from 2000 to 2009 n = 4935 n = 24098 27.5% female 28.9% female Stapley et al. (10) Pancreatic Case–control All patients in the CPRD database diagnosed with pancreatic cancer from 2000 to 2009 n = 3635 n = 16459 52.0% female 52.0% female Hamilton et al. (30) Ovarian Case–control All women in Exeter and mid- and east-Devon diagnosed with ovarian cancer from 2000 to 2007, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 39 general practices in Exeter, mid-Devon and east-Devon, UK n = 212 n = 1060 100% female 100% female Hamilton et al. (31) Colorectal Case–control All patients in Exeter, Devon, diagnosed with colorectal cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 349 n = 1744 49.3% female 49.3% female Hamilton et al. (5) Lung Case–control All patients in Exeter, Devon, diagnosed with lung cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 247 n = 1235 31.9% female 31.9% female Reference . Cancer site . Study type . Data source . Cases . Controls . Walker et al. (33) Breast Case–control All women in the CPRD database diagnosed with breast cancer from 2000 to 2009 n = 4407 n = 21755 100% female 100% female Shephard et al. (6) Kidney Case–control All patients in the CPRD database diagnosed with kidney cancer from 2000 to 2009 n = 3183 n = 15707 38.7% female 40.2% female Stapley et al. (7) Oesophago-gastric Case–control All patients in the CPRD database diagnosed with oesophago-gastric cancer from 2000 to 2009 n = 7657 n = 37699 34.6% female 34.6% female Walker et al. (8) Uterine Case–control All women in the CPRD database diagnosed with uterine cancer from 2000 to 2009 n = 3166 n = 9537 100% female 100% female Shephard et al. (32) Bladder Case–control All patients in the CPRD database diagnosed with bladder cancer from 2000 to 2009 n = 4935 n = 24098 27.5% female 28.9% female Stapley et al. (10) Pancreatic Case–control All patients in the CPRD database diagnosed with pancreatic cancer from 2000 to 2009 n = 3635 n = 16459 52.0% female 52.0% female Hamilton et al. (30) Ovarian Case–control All women in Exeter and mid- and east-Devon diagnosed with ovarian cancer from 2000 to 2007, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 39 general practices in Exeter, mid-Devon and east-Devon, UK n = 212 n = 1060 100% female 100% female Hamilton et al. (31) Colorectal Case–control All patients in Exeter, Devon, diagnosed with colorectal cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 349 n = 1744 49.3% female 49.3% female Hamilton et al. (5) Lung Case–control All patients in Exeter, Devon, diagnosed with lung cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 247 n = 1235 31.9% female 31.9% female CPRD, Clinical Practice Research Datalink. Open in new tab Table 2. Characteristics of included studies, including study type, cancer site, source of data used in the study, the number of cases and controls in each study and the proportion of females Reference . Cancer site . Study type . Data source . Cases . Controls . Walker et al. (33) Breast Case–control All women in the CPRD database diagnosed with breast cancer from 2000 to 2009 n = 4407 n = 21755 100% female 100% female Shephard et al. (6) Kidney Case–control All patients in the CPRD database diagnosed with kidney cancer from 2000 to 2009 n = 3183 n = 15707 38.7% female 40.2% female Stapley et al. (7) Oesophago-gastric Case–control All patients in the CPRD database diagnosed with oesophago-gastric cancer from 2000 to 2009 n = 7657 n = 37699 34.6% female 34.6% female Walker et al. (8) Uterine Case–control All women in the CPRD database diagnosed with uterine cancer from 2000 to 2009 n = 3166 n = 9537 100% female 100% female Shephard et al. (32) Bladder Case–control All patients in the CPRD database diagnosed with bladder cancer from 2000 to 2009 n = 4935 n = 24098 27.5% female 28.9% female Stapley et al. (10) Pancreatic Case–control All patients in the CPRD database diagnosed with pancreatic cancer from 2000 to 2009 n = 3635 n = 16459 52.0% female 52.0% female Hamilton et al. (30) Ovarian Case–control All women in Exeter and mid- and east-Devon diagnosed with ovarian cancer from 2000 to 2007, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 39 general practices in Exeter, mid-Devon and east-Devon, UK n = 212 n = 1060 100% female 100% female Hamilton et al. (31) Colorectal Case–control All patients in Exeter, Devon, diagnosed with colorectal cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 349 n = 1744 49.3% female 49.3% female Hamilton et al. (5) Lung Case–control All patients in Exeter, Devon, diagnosed with lung cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 247 n = 1235 31.9% female 31.9% female Reference . Cancer site . Study type . Data source . Cases . Controls . Walker et al. (33) Breast Case–control All women in the CPRD database diagnosed with breast cancer from 2000 to 2009 n = 4407 n = 21755 100% female 100% female Shephard et al. (6) Kidney Case–control All patients in the CPRD database diagnosed with kidney cancer from 2000 to 2009 n = 3183 n = 15707 38.7% female 40.2% female Stapley et al. (7) Oesophago-gastric Case–control All patients in the CPRD database diagnosed with oesophago-gastric cancer from 2000 to 2009 n = 7657 n = 37699 34.6% female 34.6% female Walker et al. (8) Uterine Case–control All women in the CPRD database diagnosed with uterine cancer from 2000 to 2009 n = 3166 n = 9537 100% female 100% female Shephard et al. (32) Bladder Case–control All patients in the CPRD database diagnosed with bladder cancer from 2000 to 2009 n = 4935 n = 24098 27.5% female 28.9% female Stapley et al. (10) Pancreatic Case–control All patients in the CPRD database diagnosed with pancreatic cancer from 2000 to 2009 n = 3635 n = 16459 52.0% female 52.0% female Hamilton et al. (30) Ovarian Case–control All women in Exeter and mid- and east-Devon diagnosed with ovarian cancer from 2000 to 2007, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 39 general practices in Exeter, mid-Devon and east-Devon, UK n = 212 n = 1060 100% female 100% female Hamilton et al. (31) Colorectal Case–control All patients in Exeter, Devon, diagnosed with colorectal cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 349 n = 1744 49.3% female 49.3% female Hamilton et al. (5) Lung Case–control All patients in Exeter, Devon, diagnosed with lung cancer from 1998 to 2002, identified from the Royal Devon and Exeter Hospital Cancer Registry, and electronic records of all patients at 21 general practices in Exeter, UK n = 247 n = 1235 31.9% female 31.9% female CPRD, Clinical Practice Research Datalink. Open in new tab All studies excluded patients with metastatic cancer from an earlier primary site (although included patients with metastatic disease at the time of diagnosis), patients for whom no matching controls were available, patients with no data available in the year prior to diagnosis, duplicates or controls whose matching case had been excluded. Uterine cancer controls with a hysterectomy prior to the cancer diagnosis date of their associated cases (8) and patients with a mastectomy >3 months prior to their diagnosis date (33) were excluded. All studies defined thrombocytosis as a platelet count above 400 or 450×109/l depending on the local laboratory definition. All studies had included adults aged ≥40 years. Association between thrombocytosis and cancer Four studies had carried out multivariable analyses and found a statistically significant association (P < 0.05) between thrombocytosis and their specific cancer site (5–8). The adjusted odds ratios (ORs) as presented in the publications were as follows: for lung cancer, 9.3 [95% confidence interval (CI) 3.4–26]; for kidney cancer, 2.2 (95% CI 1.7–2.7); for oesophago-gastric cancer 2.4 (95% CI 2.0–2.9) and for uterine cancer 1.50 (95% CI 1.00–2.25). Four of the remaining five studies reported no significant association between thrombocytosis and cancer (10, 30, 31, 33); these were the studies identified through contact with experts as this non-significant association was not reported in the publication The fifth study collected but did not analyse platelet count data (32). The independently calculated LRs for each cancer site are presented in Figure 2; all with the exception of breast cancer indicate an increased probability of malignancy in patients with a blood test showing thrombocytosis. Figure 2. Open in new tabDownload slide The number of cases and controls included in each study, and the number (n) of these with thrombocytosis in the year prior to diagnosis. Independently calculated likelihood ratios, with 95% confidence intervals, are estimated, and these are shown graphically in a forest plot. Figure 2. Open in new tabDownload slide The number of cases and controls included in each study, and the number (n) of these with thrombocytosis in the year prior to diagnosis. Independently calculated likelihood ratios, with 95% confidence intervals, are estimated, and these are shown graphically in a forest plot. The PPVs for each cancer site are shown in Table 3. LRs were highest in ovarian, lung, kidney, colorectal and oesophago-gastric cancer, but due to differences in the prevalence of these cancers in the consulting population, lung and colorectal (the most common cancers of the nine) had the highest PPVs for cancer. Ovarian cancer, with the greatest LR of 14.61, had a relatively small PPV of 0.65 which reflects the relative rarity of this diagnosis. Table 3. Analysis of data from included studies. The table details the number of cases and controls in each study and how many of these had a blood test result showing thrombocytosis within the year prior to diagnosis. Independently calculated positive predictive values (PPVs) are presented along with 95% confidence intervals (CI). Reference . Study source . Cancer site . Cases n, n (%) with thrombocytosis . Controls n, n (%) with thrombocytosis . Positive predictive values % (95% CI) . Hamilton et al. (5) Literature search Lung 247 1235 1.63 34 (13.8) 19 (1.5) (0.92–2.90) Shephard et al. (6) Literature search Kidney 3183 15707 0.17 348 (10.9) 251 (1.6) (0.15–0.20) Stapley et al. (7) Literature search Oesophago-gastric 7657 37699 0.47 707 (9.2) 568 (1.5) (0.42–0.52) Walker et al. (8) Literature search Uterine 3166 9537 0.08 110 (3.5) 207 (2.2) (0.07–0.11) Walker et al. (33) Contact with experts Breast 4407 21755 0.38 91 (2.1) 369 (1.7) (0.31–0.48) Shephard et al. (32) Contact with experts Bladder 4935 24098 0.10 156 (3.2) 247 (1.0) (0.08–0.12) Stapley et al. (10) Contact with experts Pancreatic 3635 16459 0.13 214 (5.9) 222 (1.3) (0.11–0.15) Hamilton et al. (30) Contact with experts Ovarian 212 1060 0.65 26 (12.3) 9 (0.8) (0.31–1.36) Stapley et al. (11) Contact with experts Colorectal 349 1744 1.39 48 (13.8) 42 (2.4) (0.94–2.09) Reference . Study source . Cancer site . Cases n, n (%) with thrombocytosis . Controls n, n (%) with thrombocytosis . Positive predictive values % (95% CI) . Hamilton et al. (5) Literature search Lung 247 1235 1.63 34 (13.8) 19 (1.5) (0.92–2.90) Shephard et al. (6) Literature search Kidney 3183 15707 0.17 348 (10.9) 251 (1.6) (0.15–0.20) Stapley et al. (7) Literature search Oesophago-gastric 7657 37699 0.47 707 (9.2) 568 (1.5) (0.42–0.52) Walker et al. (8) Literature search Uterine 3166 9537 0.08 110 (3.5) 207 (2.2) (0.07–0.11) Walker et al. (33) Contact with experts Breast 4407 21755 0.38 91 (2.1) 369 (1.7) (0.31–0.48) Shephard et al. (32) Contact with experts Bladder 4935 24098 0.10 156 (3.2) 247 (1.0) (0.08–0.12) Stapley et al. (10) Contact with experts Pancreatic 3635 16459 0.13 214 (5.9) 222 (1.3) (0.11–0.15) Hamilton et al. (30) Contact with experts Ovarian 212 1060 0.65 26 (12.3) 9 (0.8) (0.31–1.36) Stapley et al. (11) Contact with experts Colorectal 349 1744 1.39 48 (13.8) 42 (2.4) (0.94–2.09) Open in new tab Table 3. Analysis of data from included studies. The table details the number of cases and controls in each study and how many of these had a blood test result showing thrombocytosis within the year prior to diagnosis. Independently calculated positive predictive values (PPVs) are presented along with 95% confidence intervals (CI). Reference . Study source . Cancer site . Cases n, n (%) with thrombocytosis . Controls n, n (%) with thrombocytosis . Positive predictive values % (95% CI) . Hamilton et al. (5) Literature search Lung 247 1235 1.63 34 (13.8) 19 (1.5) (0.92–2.90) Shephard et al. (6) Literature search Kidney 3183 15707 0.17 348 (10.9) 251 (1.6) (0.15–0.20) Stapley et al. (7) Literature search Oesophago-gastric 7657 37699 0.47 707 (9.2) 568 (1.5) (0.42–0.52) Walker et al. (8) Literature search Uterine 3166 9537 0.08 110 (3.5) 207 (2.2) (0.07–0.11) Walker et al. (33) Contact with experts Breast 4407 21755 0.38 91 (2.1) 369 (1.7) (0.31–0.48) Shephard et al. (32) Contact with experts Bladder 4935 24098 0.10 156 (3.2) 247 (1.0) (0.08–0.12) Stapley et al. (10) Contact with experts Pancreatic 3635 16459 0.13 214 (5.9) 222 (1.3) (0.11–0.15) Hamilton et al. (30) Contact with experts Ovarian 212 1060 0.65 26 (12.3) 9 (0.8) (0.31–1.36) Stapley et al. (11) Contact with experts Colorectal 349 1744 1.39 48 (13.8) 42 (2.4) (0.94–2.09) Reference . Study source . Cancer site . Cases n, n (%) with thrombocytosis . Controls n, n (%) with thrombocytosis . Positive predictive values % (95% CI) . Hamilton et al. (5) Literature search Lung 247 1235 1.63 34 (13.8) 19 (1.5) (0.92–2.90) Shephard et al. (6) Literature search Kidney 3183 15707 0.17 348 (10.9) 251 (1.6) (0.15–0.20) Stapley et al. (7) Literature search Oesophago-gastric 7657 37699 0.47 707 (9.2) 568 (1.5) (0.42–0.52) Walker et al. (8) Literature search Uterine 3166 9537 0.08 110 (3.5) 207 (2.2) (0.07–0.11) Walker et al. (33) Contact with experts Breast 4407 21755 0.38 91 (2.1) 369 (1.7) (0.31–0.48) Shephard et al. (32) Contact with experts Bladder 4935 24098 0.10 156 (3.2) 247 (1.0) (0.08–0.12) Stapley et al. (10) Contact with experts Pancreatic 3635 16459 0.13 214 (5.9) 222 (1.3) (0.11–0.15) Hamilton et al. (30) Contact with experts Ovarian 212 1060 0.65 26 (12.3) 9 (0.8) (0.31–1.36) Stapley et al. (11) Contact with experts Colorectal 349 1744 1.39 48 (13.8) 42 (2.4) (0.94–2.09) Open in new tab Generally, studies that used CPRD data achieved larger sample sizes than those that used general practice records. The former included all patients diagnosed with the cancer of interest from any UK practice whereas the latter included patients diagnosed with the cancer of interest from set geographical areas within South-West England. Discussion Summary This is the first systematic review to identify and collate results from studies investigating the association between thrombocytosis and diagnosis of cancer in primary care. These results suggest that patients with thrombocytosis in primary care have an increased risk of cancer, and that some types of cancer are more strongly associated with thrombocytosis than others. All but one of the nine cancer sites had significant LRs calculated from raw data; for four of the nine studies, however, significance was not retained in multivariable models. Although this does not detract from the overall conclusion of an association between cancer and thrombocytosis, it does suggest that the results of this review should be interpreted with caution and supports the view that the association only exists for certain types of cancer. There did not appear to be any biological link between those cancer sites that did and did not have a significant association with thrombocytosis in multivariable models. When considering the anatomy of cancer sites, there were counterintuitive results; firstly, although colorectal and oesophago-gastric cancer, both of the digestive system, had similar ORs for cancer (6.5 and 5.9, respectively), only oesophago-gastric cancer retained significance in published multivariable models. In our independent analysis, both sites had LRs of ~5, but the PPV for colorectal cancer (1.39%) was higher than that for oesophago-gastric cancer (0.47%); this difference can be explained by the fact that colorectal cancer is more commonly diagnosed than oesophago-gastric. Secondly, uterine and ovarian cancer had somewhat disparate results. Uterine cancer alone retained significance both in the multivariable model, and in our independent analysis (LR 14.61). Key differences in the way these types of cancer develop or manifest themselves may underlie these observations. Further investigation of this could provide evidence for the biological mechanisms that underlie the association between thrombocytosis and cancer. Strengths and limitations The quality of studies included in this review was judged to be high, and the majority used CPRD data. A long-established high-quality data source, the CPRD holds primary care records for patients from 684 GP practices, covering 8.8% of the UK population with an even geographical distribution. This largely representative sample can yield studies with widely applicable results. It is particularly relevant that the study data are taken from primary care as it is in this environment that the study results should be clinically useful. The validity of CPRD data has been found to be high in two recent systematic reviews (35,36), although none of the studies included in these reviews addressed cancer data specifically. The CPRD studies include all patients in the UK within CPRD-registered practices who were diagnosed with the cancer of choice rather than patients within a restricted geographical area, as in the non-CPRD studies. This reduces the element of selection bias which is inherent in many case–control studies. It also enables much larger studies, and consequently more accurate and reliable results. However, CRPD cases are based on electronic records of cancer diagnoses alone; whereas in the non-CPRD studies, electronic records from the Cancer Registry were used as well as articles general practice records which included histology reports and were therefore more likely to be accurate. In terms of the exposure variable, the platelet counts recorded in the CPRD are electronic and automatically transmitted from laboratory to the patient records, reducing the chance of human error and subsequent risk of classification bias. Where platelet count data were missing for patients in CPRD studies, patients were assumed to not have thrombocytosis. In contrast, the non-CPRD studies relied on manual checking and recording of platelet counts, more open to sources of bias and error. Overall, the CPRD studies can be considered higher quality than the non-CPRD studies; these show a positive association between thrombocytosis and cancer, although there is marked variation between different cancer sites. Although these results can readily be applied to UK general practice, the lack of any non-UK studies limits the extent to which these findings can be generalized outside the UK. There is also a chance that relevant data may not have been identified due to publication bias. The majority of studies that contributed raw data to this study were not identified by the literature review because they did not report non-significant findings. Although the lead author of this systematic review contacted relevant experts who have carried out similar studies to ask if they had collected platelet count data (they had not), we were only able to contact those we were aware of through networking, conference attendance and general knowledge of the field. It is possible that other researchers unknown to this team are carrying out similar work and have collected data, but that this has remained unpublished due to a lack of significant findings. Another potential source of bias lies in the fact that WH, an author of this article, is an author in all nine studies included in this systematic review, and ES of three. However, the search strategy and quality assessment were carried out independently by SB and OU, who were not involved in any of the nine included studies. Comparison with existing literature The biological processes behind the thrombocytosis–cancer association have been studied in patients post-diagnosis (25,26) but the effect is not fully understood. It is also uncertain whether the mechanisms behind the association in secondary care would apply in primary care, before diagnosis is made. The association between uterine cancer and thrombocytosis is supported by a study from Stone et al. (2012) which found that 31% of patients with ovarian cancer had thrombocytosis at the point of diagnosis (38). Implications for research and practice This systematic review suggests that thrombocytosis is an early marker of some cancers in primary care and will raise awareness of this marker among GPs. This finding can be of use in primary care for GPs receiving blood results unexpectedly showing high platelet counts. Simply adding the PPVs from Table 3 suggests that the PPV from thrombocytosis is at least 5% for any of the nine cancers (several cancer sites have not been reported and any association with thrombocytosis with them would increase this figure). It would be useful to study thrombocytosis across cancer as a whole to give a more accurate estimate of the risk, but it is very likely to remain an important feature, especially now the UK has moved to rapid investigation of any cancer risk exceeding 3% (37). Any further study should also determine which specific cancers present with raised platelets as an early marker allowing targeted investigation. Crucially, this review does not suggest that platelet count should be specifically used as a diagnostic test for cancer, or as a screening tool. The current body of evidence only supports clinicians considering cancer as a diagnosis if a blood test result shows thrombocytosis. Future studies should compare risk for specific patient subgroups, including men and women, those in different age groups, and at various levels of elevated platelets. Ideally, studies could suggest the optimal platelet count to distinguish between those who do and do not have cancer. It may also be possible to combine thrombocytosis with other early markers of cancer to develop cancer-specific risk values for combinations of symptoms. Declaration Funding: the Policy Research Unit in Cancer Awareness, Screening and Early Diagnosis receives funding for a research programme from the Department of Health Policy Research Programme. It is collaboration between researchers from seven institutions (Queen Mary University of London, UCL, King’s College London, London School of Hygiene and Tropical Medicine, Hull York Medical School, Durham University and University of Exeter). OU is funded, and WH is part-funded by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care (CLAHRC) for the South West Peninsula at the Royal Devon and Exeter NHS Foundation Trust. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health in England. Ethical approval: none. Conflict of interest: none. Acknowledgements Many thanks to Sal Stapley and Sarah Walker for sharing data from their studies. Thanks also to Tim Starkey for producing the forest plot in Matlab. References 1. Coleman MP Forman D Bryant H et al. Cancer survival in Australia, Canada, Denmark, Norway, Sweden, and the UK, 1995–2007 (the International Cancer Benchmarking Partnership): an analysis of population-based cancer registry data . Lancet 2011 ; 377 : 127 – 38 . Google Scholar Crossref Search ADS PubMed WorldCat 2. Richards MA . The size of the prize for earlier diagnosis of cancer in England . Br J Cancer 2009 ; 101 : S125 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat 3. Hamilton W . Cancer diagnosis in primary care . Br J Gen Pract 2010 ; 60 : 121 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 4. NICE . Referral Guidelines for Suspected Cancer . London, UK : NICE , 2005 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 5. Hamilton W Peters TJ Round A Sharp D . What are the clinical features of lung cancer before the diagnosis is made? A population based case-control study . Thorax 2005 ; 60 : 1059 – 65 . Google Scholar Crossref Search ADS PubMed WorldCat 6. Shephard E Neal R Rose P et al. Clinical features of kidney cancer in primary care: a case-control study using primary care records . Br J Gen Pract 2013 ; 63 : e250 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat 7. Stapley S Peters TJ Neal RD et al. The risk of oesophago-gastric cancer in symptomatic patients in primary care: a large case-control study using electronic records . Br J Cancer 2013 ; 108 : 25 – 31 . Google Scholar Crossref Search ADS PubMed WorldCat 8. Walker S Hyde C Hamilton W . Risk of uterine cancer in symptomatic women in primary care: case-control study using electronic records . Br J Gen Pract 2013 ; 63 : e643 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 9. Barrett J Sharp DJ Stapley S et al. Pathways to the diagnosis of ovarian cancer in the UK: a cohort study in primary care . BJOG 2010 ; 117 : 610 – 4 . Google Scholar Crossref Search ADS PubMed WorldCat 10. Stapley S Peters TJ Neal RD et al. The risk of pancreatic cancer in symptomatic patients in primary care: a large case-control study using electronic records . Br J Cancer 2012 ; 106 : 1940 – 4 . Google Scholar Crossref Search ADS PubMed WorldCat 11. Stapley S Peters TJ Sharp D Hamilton W . The mortality of colorectal cancer in relation to the initial symptom at presentation to primary care and to the duration of symptoms: a cohort study using medical records . Br J Cancer 2006 ; 95 : 1321 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat 12. Aoe K Hiraki A Ueoka H et al. Thrombocytosis as a useful prognostic indicator in patients with lung cancer . Respiration 2004 ; 71 : 170 – 3 . Google Scholar Crossref Search ADS PubMed WorldCat 13. Bensalah K Leray E Fergelot P et al. Prognostic value of thrombocytosis in renal cell carcinoma . J Urol 2006 ; 175 : 859 – 63 . Google Scholar Crossref Search ADS PubMed WorldCat 14. Carr BI Guerra V . Thrombocytosis and hepatocellular carcinoma . Dig Dis Sci 2013 ; 58 : 1790 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat 15. Cravioto-Villanueva A Luna-Perez P Gutierrez-de la Barrera M et al. Thrombocytosis as a predictor of distant recurrence in patients with rectal cancer . Arch Med Res 2012 ; 43 : 305 – 11 . Google Scholar Crossref Search ADS PubMed WorldCat 16. Gogus C Baltaci S Filiz E et al. Significance of thrombocytosis for determining prognosis in patients with localized renal cell carcinoma . Urology 2004 ; 63 : 447 – 50 . Google Scholar Crossref Search ADS PubMed WorldCat 17. Gucer F Tamussino K Keil F et al. Thrombocytosis in gynecologic malignancies . Anticancer Res 2004 ; 24 : 2053 – 9 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 18. Holgersson G Sandelin M Hoye E et al. Swedish lung cancer radiation study group: the prognostic value of anaemia, thrombocytosis and leukocytosis at time of diagnosis in patients with non-small cell lung cancer . Med Oncol 2012 ; 29 : 3176 – 82 . Google Scholar Crossref Search ADS PubMed WorldCat 19. Hwang SG Kim KM Cheong JH et al. . Impact of pretreatment thrombocytosis on blood-borne metastasis and prognosis of gastric cancer . Eur J Surg Oncol 2012 ; 38 : 562 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat 20. Inoue K Kohashikawa K Suzuki S et al. Prognostic significance of thrombocytosis in renal cell carcinoma patients . Int J Urol 2004 ; 11 : 364 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat 21. Ishizuka M Nagata H Takagi K et al. Preoperative thrombocytosis is associated with survival after surgery for colorectal cancer . J Surg Oncol 2012 ; 106 : 887 – 91 . Google Scholar Crossref Search ADS PubMed WorldCat 22. Li AJ Madden AC Cass I et al. The prognostic significance of thrombocytosis in epithelial ovarian carcinoma . Gynecol Oncol 2004 ; 92 : 211 – 4 . Google Scholar Crossref Search ADS PubMed WorldCat 23. Pedersen LM Milman N . Prognostic significance of thrombocytosis in patients with primary lung cancer . Eur Respir J 1996 ; 9 : 1826 – 30 . Google Scholar Crossref Search ADS PubMed WorldCat 24. Suppiah R Shaheen PE Elson P et al. Thrombocytosis as a prognostic factor for survival in patients with metastatic renal cell carcinoma . Cancer 2006 ; 107 : 1793 – 800 . Google Scholar Crossref Search ADS PubMed WorldCat 25. Arslan Ç Coşkun HŞ . Thrombocytosis in solid tumors: review of the literature . Turk J Haematol 2005 ; 22 : 59 – 64 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 26. Nash GF Turner LF Scully MF Kakkar AK . Platelets and cancer . Lancet Oncol 2002 ; 3 : 425 – 30 . Google Scholar Crossref Search ADS PubMed WorldCat 27. Whiting PF Rutjes AW Westwood ME et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies . Ann Intern Med 2011 ; 155 : 529 – 36 . Google Scholar Crossref Search ADS PubMed WorldCat 28. Centre for Reviews and Dissemination . Systematic Reviews: CRD’s Guidance for Undertaking Reviews in Health Care . York, UK : CRD, University of York , 2008 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 29. Knottnerus JA. The Evidence Base of Clinical Diagnosis . London, UK : BMJ Books , 2002 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 30. Hamilton W Peters TJ Bankhead C Sharp D . Risk of ovarian cancer in women with symptoms in primary care: population based case-control study . Br Med J 2009 ; 339 : b2998 . Google Scholar Crossref Search ADS WorldCat 31. Hamilton W Round A Sharp D Peters TJ . Clinical features of colorectal cancer before diagnosis: a population-based case-control study . Br J Cancer 2005 ; 93 : 399 – 405 . Google Scholar Crossref Search ADS PubMed WorldCat 32. Shephard E Stapley S Neal RD et al. Clinical features of bladder cancer in primary care . Br J Gen Pract 2012 ; 62 : e598 – 604 . Google Scholar Crossref Search ADS PubMed WorldCat 33. Walker S Hyde C Hamilton W . Risk of breast cancer in symptomatic women in primary care: a case control study using primary care records . Br J Gen Pract 2014 ; 64 : e788 – 93 . Google Scholar Crossref Search ADS PubMed WorldCat 34. Williams T van Staa T Puri S Eaton S . Recent advances in the utility and use of the general practice research database as an example of a UK primary care data resource . Ther Adv Drug Saf 2012 ; 3 : 89 – 99 . Google Scholar Crossref Search ADS PubMed WorldCat 35. Herrett E Thomas SL Schoonen WM et al. Validation and validity of diagnoses in the general practice research database: a systematic review . Br J Clin Pharmacol 2010 ; 69 : 4 – 14 . Google Scholar Crossref Search ADS PubMed WorldCat 36. Khan NF Harrison SE Rose PW . Validity of diagnostic coding within the general practice research database: a systematic review . Br J Gen Pract 2010 ; 60 : e128 – 36 . Google Scholar Crossref Search ADS PubMed WorldCat 37. NICE . Suspected Cancer: Recognition and Referral . London, UK : NICE , 2015 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 38. Stone R Nick A McNeish I et al. Paraneoplastic thrombocytosis in ovarian cancer . NEJM 2012 ; 366 : 610 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Author notes " *Correspondence to Sarah Bailey, Primary Care Diagnostics, University of Exeter Medical School, College House, St Luke’s Campus, University of Exeter, Exeter, Devon EX1 2LU, UK; E-mail: s.e.r.bailey@exeter.ac.uk © The Author 2016. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

Journal

Family PracticeOxford University Press

Published: Feb 1, 2017

There are no references for this article.