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

Learn More →

Treatment of gastrointestinal tumor (GIST) of the rectum requiring abdominoperineal resection following neoadjuvant imatinib: a cost-effectiveness analysis

Treatment of gastrointestinal tumor (GIST) of the rectum requiring abdominoperineal resection... Background: Neoadjuvant imatinib for gastrointestinal stromal tumors (GIST ) of the rectum can reduce, but may not eliminate, risk of surgical morbidity from permanent bowel diversion. We sought to evaluate the cost-effectiveness of alternative strategies in rectal GIST patients requiring abdominoperineal resection following neoadjuvant imatinib. Methods: We developed a Markov model using a healthcare payers’ perspective to estimate costs in 2017 Singapore dollars (SGD) and quality adjusted life years (QALYs) for upfront abdominoperineal resection (UAPR) versus continued imatinib until progression (CIUP) following 1 year of neoadjuvant imatinib. Transition probabilities and utilities were obtained from published data, and costs were estimated using data from the National Cancer Centre Singapore. Deterministic and probabilistic sensitivity analyses were conducted to probe model uncertainty. Incremental cost- effectiveness ratio below SGD 50,000 per QALY gained was considered cost-effective. Results: In the base case, UAPR dominates CIUP being both more effective (8.66 QALYS vs 5.43 QALYs) and less expensive (SGD 312,627 vs SGD 339,011). These estimates were most sensitive to 2 variables, utility of abdominoper- ineal resection and annual recurrence probability post-abdominoperineal resection. However, simultaneously varying the values of these variables to maximally favor CIUP did not render it the more cost effective strategy at willingness to pay ( WTP) of SGD 50,000. In probabilistic sensitivity analysis, UAPR had probability of being cost-effective com- pared with CIUP greater than 95%, reaching 100% at WTP SGD 10,000. Conclusion: UAPR is more effective and less costly than CIUP for patients with rectal GIST requiring abdominoper - ineal resection following neoadjuvant imatinib, and is the strategy of choice in this setting. Keywords: Rectal GIST treatment, Imatinib, Cost effectiveness analysis Background million per year [1]. The vast majority of cases arise in Gastrointestinal stromal tumors (GIST) are mesenchy- the stomach (40–60%), or jejunum/ileum (25–30%); mal tumors arising from the interstitial cells of Cajal approximately 5–15% of cases arise from the colon or in the alimentary tract with an incidence of 10–15 per rectum [2]. The primary treatment of GIST is com - plete surgical resection; among individuals without high-risk clinicopathologic features for recurrence this *Correspondence: Mohamad.farid@singhealth.com.sg approach results in sustained disease control in up to Division of Medical Oncology, National Cancer Centre Singapore, 11 70% of cases [3]. Imatinib, the tyrosine kinase inhibi- Hospital Drive, Singapore 169610, Singapore Full list of author information is available at the end of the article tor antagonizing the KIT and PDGFR oncoproteins © The Author(s) 2020. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/publi cdoma in/ zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Farid et al. Clin Sarcoma Res (2020) 10:13 Page 2 of 11 implicated in the majority of GISTs, has revolution- would require abdominoperineal resection for surgical ized the treatment of this previously dismal disease. In extirpation. advanced disease, imatinib achieves objective response rates, progression free survival (PFS) and overall sur- vival (OS) of 50%, 1.5 years and 4 years, respectively [4]. Methods In addition, 3 year of adjuvant imatinib improves over- Theoretical model/model overview all survival in patients with high-risk localized disease A comprehensive Markov model was developed to ana- who have undergone resection [5]. The achievement of lyze the health and cost impact of alternative strategies this significant therapeutic advance has come with the in management of rectal GIST from a healthcare sec- convenience of a pill taken once daily having an excel- tors’ perspective in Singapore. The structure of a Markov lent therapeutic index and minimal cumulative toxicity model is composed of mutually exclusive health states [6]. GISTs arising from the rectum poses a therapeutic relevant to the impact of disease and treatments under challenge in view of potential for significant surgical evaluation. Over a pre-determined time horizon, any morbidity from loss of bowel-sphincteric control and patient can stay in only 1 health state at any time, but may need for permanent bowel diversion if abdominoper- move between health states over time [14]. We modelled ineal resection is required for complete resection [7]. a hypothetical cohort of 60-year old patients with rectal Neoadjuvant imatinib has been shown to be safe and GIST who had received 1  year of neoadjuvant imatinib efficacious in the treatment of GIST presenting chal - at 400 milligrams daily with no disease progression, for lenges to upfront surgery, including disease arising which resection would necessitate an abdominoperineal from the rectum [8, 9]. Expert guidelines suggest that resection. The model evaluated 2 treatment options - neoadjuvant imatinib be given for up to 12  months continued imatinib until progression (CIUP), or surgi- in such cases to maximize response prior to surgery, cal resection with upfront abdominoperineal resection with consideration given for more imatinib postopera- (UAPR). (Figure 1) The model had a cycle length of 1 year tively to achieve a total of 3  years of imatinib therapy and followed patients over a time horizon of 20 years. A [10]. Although surgical resection remains the corner- total of 12 health states were defined: UAPR at 1st year stone of curative therapy, some patients continue to (“UAPR_Yr1”), UAPR at 2nd year (“UAPR_Yr2”) and defer curative surgery at the end of this period in view UAPR at 3rd year and beyond (“UAPR_Yr3+”) repre- of the excellent therapeutic index of imatinib and fear sent patients at different time-points following upfront of permanent morbidity from APR, the need for which abdominoperineal resection. The fourth health state may persist even after maximal tumor response to neo- accounts for patients receiving salvage surgery for their adjuvant imatinib. These patients may elect to con - 1st local recurrence following upfront abdominoperineal tinue imatinib indefinitely in lieu of upfront surgery, to resection (“Salvage Sx following LR1 post UAPR_Yr3+”). avoid surgical morbidity unless absolutely necessary, if The fifth health state addresses patients receiving the and when they progress on imatinib. Notwithstanding second strategy, CIUP (“CIUP”). The sixth and seventh the excellent therapeutic index of imatinib, there are health states contains patients following local progression no data to suggest that indefinite imatinib can substi - on CIUP–either undergoing abdominoperineal resection tute surgery as curative treatment in localized disease. following local progression on CIUP (“abdominoperineal The natural history for patients receiving imatinib in resection following LR on CIUP”) or subsequently sal- advanced disease is the eventual development of resist- vage surgery following 1st local recurrence after abdomi- ance and consequent local as well as distant progres- noperineal resection on CIUP (“Salvage Sx following LR1 sion. The development of metastatic disease precludes post CIUP”). The eighth through eleventh health states therapy with curative intent and leads to eventual encompasses patients at different lines of palliative ther - morbidity and mortality [11]. While imatinib has been apy in advanced disease, namely: distant recurrence (“1st shown to be cost-effective in both the adjuvant [12] and DR”) where they receive imatinib 400 milligrams daily advanced [13] disease setting, there are no data on the [11]; 1st progression in metastatic disease (“mets disease cost-effectiveness of imatinib as primary therapy in a 1st PD”) where they receive imatinib 800 milligrams daily setting where upfront surgery has significant potential [11]; 2nd progression in metastatic disease (“mets disease morbidity, specifically in rectal GIST. 2nd PD”) where they receive sunitinib 50 milligrams daily We sought to estimate, using a Markov decision 4 weeks on and 2 weeks off in a 6-weekly cycle [15]; and model, the relative effectiveness and cost-effectiveness 3rd progression in metastatic disease (“mets disease 3rd of upfront abdominoperineal resection compared with PD”) where they receive regorafenib 160 milligrams daily continued imatinib until progression for patients with 3  weeks on and 1  week off in a 4-weekly cycle [16]. The rectal GIST having received neoadjuvant imatinib who twelfth and final health state represents death (“Dead”). F arid et al. Clin Sarcoma Res (2020) 10:13 Page 3 of 11 Fig. 1 Markov model structure evaluating 2 treatment strategies (UAPR and CIUP) with 12 health states Farid et al. Clin Sarcoma Res (2020) 10:13 Page 4 of 11 For patients on UAPR, they received 2 years of postop- the former study much more closely resembled that used erative imatinib, to make up a total of 3 years of neoadju- in our model (12  months of imatinib at 400 milligrams vant and adjuvant imatinib, as consistent with treatment daily), as opposed to the 2 months of 600 milligrams daily guidelines [10]. Patients with disease progression while neoadjuvant imatinib employed in the RTOG 0132 study on adjuvant imatinib have a dismal prognosis, and thus [9]. In addition, there were 33 patients with rectal GIST transit directly to the health state where they receive in the retrospective study, as opposed to only 3 with rec- escalated doses of imatinib, “mets disease 1st PD”, regard- tal disease in RTOG 0132. For these reasons, we decided less of whether the progression was local or distant. The to use data obtained from the retrospective study in our first local recurrence after patients complete 2  years of model. The range of values considered for these transi - adjuvant imatinib therapy can yet be treated with salvage tion probabilities were derived from the confidence inter - surgery followed by 3 further years of adjuvant imatinib, vals reported in the literature or estimated if published a strategy with demonstrated benefit in newly diagnosed data were unavailable (Table 1). high-risk resected GIST [5]. The second local recurrence The model simulated transitions between health states or distant recurrence would lead to patients entering the with a cycle length of 1 year, which was chosen as a clini- metastatic disease health state receiving first line pallia - cally and therapeutically relevant interval for evaluation, tive therapy with imatinib, ‘1st DR’. considering the duration of GIST neoadjuvant and adju- For patients on CIUP, local progression would be vant (1 and 2  years, respectively) treatment and average treated with abdominoperineal resection followed by duration of disease control. The probability of transition - no further Imatinib. The first local recurrence following ing to death during each cycle was defined as the maxi - such an abdominoperineal resection would be treated mum value of observed mortality rate by using survival with salvage surgery only. The second local recurrence or data from studies in GIST as well as the background distant recurrence would lead to patients directly enter- mortality rate, estimated from age-specific death rates in ing the advanced disease health state receiving esca- Singapore [20]. The mortality of abdominoperineal resec - lated doses of imatinib (“mets disease 1st PD”), as these tion and subsequent salvage surgery for first post-abdom - patients have demonstrated failure to treatment with inoperineal resection local recurrence was assumed to be imatinib 400 milligrams daily. Tunnels states were used 0.5%. to facilitate the undertaking of salvage surgery in the set- ting of local recurrence (following APR in both the APR Utilities and costs and IUP arms) only once. QALYs were calculated by multiplying the time spent in a Model outcomes were defined as treatment costs (in given state (in life-years) by the utility score (a health sta- 2017 SGD) and quality-adjusted life years (QALYs). tus value ranging from 0 for death to 1 for perfect health) We calculated the incremental cost-effectiveness ratios associated with that state. The utility weights of all health (ICERs) of the more expensive to the less expensive states were derived from published studies (Table  1). strategy as the difference in costs divided by the differ - Utility associated with abdominoperineal resection was ence in QALYs – to compare 2 treatment strategies. All obtained from studies performed in recurrent rectal can- costs and health outcomes were discounted by 3% annu- cer [7]. Costs of surgery were estimated based on data ally. We adopted a willingness to pay (WTP) and ICER of from the National Cancer Centre Singapore (NCCS). SGD 50,000 per QALY gained as the threshold for cost- Drug costs were derived from the NCCS pharmacy data. effectiveness [17]. The model was developed using the Clinical consultation, blood tests and computed tomog- decision analytic software TreeAge Pro 2017 (TreeAge raphy (CT) scan performed every 3 monthly as surveil- Software, Williamstown, MA). lance were estimated to cost 3000 SGD per year based on expert opinion and NCCS data. Model transitions and survival estimates Event rates for progression and mortality were used to calculate transition probabilities [18, 19]. For advanced Sensitivity analysis GIST, data was obtained from randomized controlled To evaluate the robustness of the model and address the trials in the first [4], second [15] and third [16] line set - uncertainty in estimation of variables, we performed ting for GIST treatment. Recurrence rates following neo- a series of one-way deterministic sensitivity analyses adjuvant imatinib and surgery for localized GIST were (DSA), in each instance varying the value of one param- obtained from a large retrospective study of neoadjuvant eter at a time over its defined range and examining the imatinib in GIST [8]. Although there has been a prospec- effect of each parameter individually on ICERs for all tive single arm phase 2 study of neoadjuvant imatinib in variables. Where available, the range of values consid- localized GIST [9], the perioperative imatinib regimen in ered for these sensitivity analyses were derived from F arid et al. Clin Sarcoma Res (2020) 10:13 Page 5 of 11 Table 1 Transition probabilities, utilities and costs Variable Base case Range Distribution Reference for sensitivity for probabilistic analysis sensitivity analysis Probabilities Annual probability of recurrence post abdominoperineal resection 0.0871 0.0435–0.130 Beta Rutkowski et al. [8] Annual conditional probability of local recurrence post abdominoper- 0.135 0.0675–0.203 Uniform Rutkowski et al. [8] ineal resection Annual probability of 1st progression in metastatic GIST 0.370 0.296–0.444 Beta Blanke et al. [4] Annual probability of 2nd progression in metastatic GIST 0.811 0.649–0.973 Beta Blanke et al. [4] Annual probability of 3rd progression in metastatic GIST 0.708 0.566–0.850 Beta Demetri et al. [13] Annual probability of death in metastatic GIST post-regorafenib 0.405 0.270–0.410 Beta Demetri et al. [14] Utilities Recurrence-free health state post abdominoperineal resection 0.830 0.650–1 Beta Miller et al. [7] Recurrence-free health state on continued imatinib until progression 0.935 0.750–1 Beta Wilson et al. [17] GIST recurrence 0.748 0.598–0.898 Beta Majer et al. [10] GIST 1st progression in metastatic disease 0.712 0.685–0.739 Beta Chabot et al. [18] GIST 2nd progression in metastatic disease 0.712 0.685–0.739 Beta Assumption GIST 3rd progression in metastatic disease 0.712 0.685–0.739 Beta Assumption COSTS (SGD) Annual cost of imatinib 400 mg once daily 37 040 7 408–44 448 Gamma NCCS data Annual cost of sunitinib 50 mg once daily 4 weeks on, 2 weeks off 64 063 51 250–76 876 Gamma NCCS data Annual cost of regorafenib 160 mg once daily 3 weeks on, 1 week off 72 001 57 601–86 401 Gamma NCCS data Abdominoperineal resection 38 000 30 400 –45 600 Gamma NCCS data Salvage surgery (following 1st local recurrence post abdominoperineal 38 000 30 400–45 600 Gamma NCCS data resection) Annual cost of follow-up (consultation, blood tests, computer tomog- 3 000 2 400–3 600 Gamma NCCS data raphy scans every 3 months) mg milligrams, GIST gastrointestinal stromal tumor, NCCS National Cancer Centre Singapore, SGD Singapore dollars ± 50% (wider intervals used in estimation due to paucity of systematic data for recurrences specifically in rectal GIST post neoadjuvant imatinib) ± 20% + 20% and −80% (lower bound for imatinib extended to −80% to account for possible significant decrease in imatinib cost with advent of generic imatinib) the confidence intervals reported in the primary litera - Results ture (Table 1). Base case/main analysis To account for variation in multiple parameters Examining the 2 treatment strategies over the 20-year simultaneously, we also completed probabilistic sen- time horizon, UAPR (SGD 312,627) is less costly com- sitivity analyses (PSA). We performed 10,000 Monte pared to CIUP (SGD 339,011), yet generates more QALYs Carlo simulations, each time randomly sampling from (8.66 vs 5.43  years), thus dominating over CIUP as a the distributions for all parameters simultaneously. strategy. Uncertainty in utilities was represented by β distribu- tions which are bounded by 0 and 1, while uncertainty in cost was represented by γ distributions, which are Sensitivity analysis bounded by 0 and infinity [21]. To succinctly represent One-way DSA revealed our results to be robust for all uncertainty surrounding our base case estimate of cost- variables across the pre-defined range of values (Table  1), effectiveness, cost-effectiveness acceptability curves with UAPR consistently dominating over CIUP; the only were derived and used to project the probability that exception to this arose in the context of annual recur- each treatment strategy was economically preferred rence probability post abdominoperineal resection under various WTP thresholds. greater than 12.5%, where an ICER for UAPR emerged, Farid et al. Clin Sarcoma Res (2020) 10:13 Page 6 of 11 Fig. 2 One way sensitivity analysis evaluating incremental cost effectiveness ratios (ICERs) of UAPR compared with CIUP for a range of utilities associated with abdominoperineal resection though remaining well below SGD 50,000 per QALY The probability of UAPR being cost-effective compared gained. with CIUP is 100% for WTP SGD 10,000 and above, and We performed several scenario analyses to further never drops below 95%. Consistent with the findings delineate the impact of extreme values for several vari- from the base case analysis, the cost-effectiveness plane ables on our outcome. When considering extremely low reveals UAPR to be dominant over CIUP for the majority values of utility for the abdominoperineal resection, of iterations of the PSA at a WTP of SGD 50,000. CIUP generated more QALYs than UAPR for utility val- ues below 0.34, with CIUP producing ICER less than SGD Discussion 50,000 for utility values less than 0.24 (Fig. 2 ). As earlier In this cost-effectiveness analysis, considering rec - described, UAPR ceased to dominate CIUP when post- tal GIST patients with no disease progression follow- abdominoperineal resection recurrence exceeded 12%. ing 12  months of neoadjuvant imatinib who required Extending this analysis to consider even higher annual abdominoperineal resection for surgical extirpation, probabilities of recurrence post abdominoperineal resec- UAPR is more effective and less costly than CIUP. These tion, UAPR generated more QALYs for annual recurrence results remained robust after considering one-way DSA probabilities below 47.0%; for higher annual recurrence of multiple factors. The only exception to this arose when probabilities, CIUP became the dominant strategy. UAPR the annual probability of recurrence following abdomi- remained more cost-effective than CIUP at WTP of SGD noperineal resection exceeded 12.5%; in this situation, 50,000 for annual recurrence probabilities below 31% UAPR was costlier and more effective than CIUP, with (Fig.  3). Additionally, we performed a 2-way sensitivity ICER values remaining well below the SGD 50,000 WTP analysis exploring extreme values of utility for abdomin- threshold. These findings were replicated in PSA, with operineal resection (as low as 0.5) and annual recurrence the probability of UAPR being cost-effective compared probabilities post abdominoperineal resection (as high as with CIUP being 100% from a WTP threshold of SGD 35%). In this analysis, the net monetary benefit for UAPR 10,000 onwards. Taken in aggregate, our data suggest would always be superior for annual recurrence probabil- that, consistent with current practice, surgical extirpation ities lower than 18% (Fig. 4). must necessarily remain the primary curative treatment From the PSA, the cost-effectiveness acceptability modality in localized rectal GIST. Such a conclusion curve is shown in Fig. 5. would be consistent with the practice and evidence from the treatment of solid tumors in general [22], and GIST F arid et al. Clin Sarcoma Res (2020) 10:13 Page 7 of 11 Fig. 3 One way sensitivity analysis evaluating incremental cost effectiveness ratios (ICERs) of UAPR compared with CIUP for a range of annual recurrence probabilities post abdominoperineal resection. Below a recurrence probability of 12%, UAPR dominates CIUP–it costs less and is more effective, thus generating no meaningful ICER. The range of recurrence probabilities considered thus begins with 15% in particular [3]. Complete resection should thus not be even if a 50% uncertainty of this point estimate is con- forsaken in the treatment of rectal GIST. sidered as an estimate of a 95% confidence interval, the To account for the possibility of widespread individ- upper bound of this confidence interval would be 13.0%. ual variation in utility of abdominoperineal resection, Considering other data, the annual probability of recur- we performed one way scenario analyses to explore the rence for high risk GIST resected upfront followed by impact of extreme aversion to abdominoperineal resec- 3 years of adjuvant imatinib in the landmark prospective tion through evaluation of very low values of utility asso- study of adjuvant imatinib was 6.9% [5], and the annual ciated with abdominoperineal resection. We found that recurrence probability of recurrence amongst imatinib- CIUP would be more cost effective than UAPR (WTP of treated rectal GIST patients in another retrospective SGD 50,000) only if utility values were 0.24 or lower. This study was 5.7% [23]. These data suggest it extremely hypothetical value, assigning extraordinarily severe disu- improbably that annual recurrence rates post resection of tility to abdominoperineal resection, is much lower than rectal GIST treated with imatinib would be high enough the value of 0.83 described in the literature [7], and may for UAPR to be rendered cost-ineffective compared with be argued to be implausible in clinical practice. Nonethe- CIUP. less, we think that this information may be useful as a Probing the sensitivity to these variables further, we quantitative threshold to guide individualized discussions simultaneously explored extreme values of both util- between physicians and patients with extreme aversion ity associated with abdominoperineal resection and to abdominoperineal resection following neoadjuvant annual recurrence probabilities post abdominoperineal imatinib for rectal GIST. resection in a 2-way sensitivity analysis. In this analysis, As earlier described, UAPR ceased to dominate CIUP UAPR was shown to be the optimal strategy at a WTP for annual recurrence probabilities following abdomi- of SGD 50,000 for all cases when post abdominoperineal noperineal resection greater than 12.5%. When more resection recurrence was less than 18% annually, regard- extreme values of recurrence probabilities were evaluated less of how low the utility associated with abdominop- in a scenario analysis, the ICER of UAPR remained below erineal resection was (considering utility values as low SGD 50,000 for recurrence probabilities lower than as 0.5). Even when considering the lower bound of the 31.5%. By comparison, the annual recurrence probability confidence interval for utility associated with abdomi - following resection after neoadjuvant imatinib is 8.7% [8]; noperineal resection (0.65) [7], UAPR would remain Farid et al. Clin Sarcoma Res (2020) 10:13 Page 8 of 11 Fig. 4 Two way sensitivity analysis comparing the net monetary benefit (NMB) of UAPR vs CIUP at a willingness to pay of SGD 50,000 when simultaneously considering varying values of utility associated with abdominoperineal resection and annual recurrence probabilities post abdominoperineal resection. Red denotes UAPR having superior NMB, while blue denotes CIUP having superior NMB more cost-effective for annual recurrence probabili - (Fig.  6). As ICER is calculated by dividing the difference ties post abdominoperineal resection below 24%. Either in costs by the difference in effectiveness, and given that way, these annual recurrence probabilities below which CIUP was less effective than UAPR, this suggests that UAPR would be the optimal strategy are well beyond the the cost of the CIUP strategy was greater than the UAPR upper limit of the confidence interval for annual recur - strategy, even if imatinib were free. This increased cost is rence probability of 13%, reinforcing the robustness likely related to the costs accruing from inferior clinical of the result favoring the superiority of UAPR. In addi- outcomes with CIUP. tion, notwithstanding the improbability of such high There are several limitations to our study. Our analy - annual recurrence rates in clinical practice, these data sis applies only to patients for whom abdominoperineal would be useful in therapeutic discussions with clinico- resection is required following neoadjuvant imatinib, a pathologically high risk rectal GIST patients expressing situation that applies to the minority of patients. In the severe aversion to the morbidity associated with abdomi- imatinib era, with the excellent clinical responses and noperineal resection. Formally quantifying individual long term disease control afforded by imatinib, there preferences, and comparing it against quantitated bound- are a range of less morbid surgical procedures avail- aries of transition probabilities associated with particular able to patients without compromising completeness of strategies, will hopefully enable more informed and pre- surgical extirpation. In one of the largest series of rec- cise therapeutic decision making. tal GIST reported, the rate of patients requiring APR The cost of imatinib was not a significant factor in or pelvic exenteration in the imatinib era was only 3%, determining the optimal strategy in this analysis. Since compared with 59% before the advent of imatinib [24]. CIUP was less effective than UAPR, CIUP would not be Nonetheless, our aim was to very specifically evaluate more cost effective even if it was the less costly strat - the outcomes for the most morbid surgical scenario— egy. As it were, when cost effectiveness of CIUP rela - loss of bowel continence—to see if deferring surgery tive to UAPR was evaluated in relation to how the cost could possibly be the more cost-effective option fol - of imatinib varied in a one-way sensitivity analysis, there lowing neoadjuvant imatinib, an option that patients was no faced with such a clinical scenario may be inclined to meaningful ICER accrued (the ICER remained nega- consider (as distinct from the situation when surgery tive) even when the cost of imatinib was zero dollars has less morbidity, and thus opting for upfront surgery F arid et al. Clin Sarcoma Res (2020) 10:13 Page 9 of 11 Fig. 5 Cost effectiveness acceptability curve from probabilistic sensitivity analysis comparing UAPR and CIUP. UAPR has a 100% probability of being more cost effective that CIUP for willingness to pay of SGD 10,000 and above Fig. 6 One way sensitivity analysis evaluating incremental cost effectiveness ratios (ICERs) of CIUP compared with UAPR for a range of costs of imatinib. For all values of imatinib price, no meaningful ICER is accrued. In the setting of the known finding of CIUP having less effectiveness, this suggests that it is the more costly strategy overall regardless of imatinib cost is more clear-cut). As it turns out, our analysis confirms therapy, primarily because the toxicity from imatinib is the importance of complete surgical extirpation even if generally limited and tolerable [25]. Additionally, drug surgical morbidity is a certainty. We did not consider toxicities from sunitinib and regorafenib would only the utilities and costs of toxicity accruing from drug become relevant in the setting of advanced disease, and Farid et al. Clin Sarcoma Res (2020) 10:13 Page 10 of 11 Authors’ contributions would not differentially affect costs and effectiveness MF and DM conceptualized and designed the study. MF, JO, CC, GT, MT, across the 2 strategies. We did not consider indirect RQ and JT collected and assembled the data. MF and DM analyzed and costs such as loss of earnings from disrupted employ- interpreted the data. All authors wrote and approved the final manuscript. All authors read and approved the final manuscript. ment or costs to unpaid caregivers, given that we employed a healthcare sectors’ perspective as opposed Funding to a societal perspective in our analysis [26]. We did Nil, there was no funding support. not include disease related prognostic factors like size, Availability of data and materials mitotic rate and mutational status in our model. This is All data generated and/or analyzed during this study are included in this because full clinicopathologic characterization (patho- published article. logical size as well as mitotic count) is not available at Ethics approval and consent to participate the start of neoadjuvant imatinib for rectal GIST, where Not applicable diagnosis is often based on endoscopically acquired Consent for publication fine needle biopsy —so these factors do not dictate ini - Not applicable. tiation of imatinib in clinical practice. In addition, the primary value of mutational status is in ensuring only Competing interests RQ has received honoraria from, served in a consulting or advisory role for, patients with imatinib sensitive mutations (KIT muta- and been part of a speakers’ bureau for Novartis, Bayer, Merck and Eisai. RQ has tions, and PDGFRA mutations other than D842V) received honoraria and served in a consulting or advisory role for Bristol-Myers receive imatinib–this group forms the vast majority Squibb (BMS). RQ has received travel, accommodation and expenses from Roche, Novartis and Eisai. RQ has received education grants from Novartis, (> 80%) of GIST patients. There were, however, patient Eisai, Janssen and Bayer. (surgical fitness, specific anatomical factors, comorbid - ities and resources for self-care, perceptions of surgical Author details Division of Medical Oncology, National Cancer Centre Singapore, 11 Hospital morbidity) and treatment (surgical expertise) prognos- Drive, Singapore 169610, Singapore. Division of Surgical Oncology, National tic factors that we did not account for. It may be argued Cancer Centre, Singapore, Singapore. Parkway Cancer Centre, Singapore, that a WTP of SGD 50,000 is too conservative for con- Singapore. Radiation Oncology, Farrer Park Hospital, Singapore, Singapore. Programme in Health Services and Systems Research, Graduate Medical temporary considerations of cost-effectiveness [17]; School, Duke-National University of Singapore, Singapore, Singapore. however, given the robustness of our results, the con- sideration of higher levels of WTP is not likely to alter Received: 13 February 2018 Accepted: 30 July 2020 our conclusions. Conclusions References In summary, our cost-effectiveness analysis confirms the 1. Soreide K, et al. Global epidemiology of gastrointestinal stromal tumours importance of surgical extirpation in rectal GIST patients (GIST ): a systematic review of population-based cohort studies. Cancer who have received neoadjuvant imatinib, showing Epidemiol. 2016;40:39–46. 2. Morgan, J., Raut CP. Epidemiology, classification, clinical presenta- upfront abdominoperineal resection to be more effective tion, prognostic features, and diagnostic work-up of gastrointestinal and less costly than continued imatinib until progression. mesenchymal neoplasms including GIST. UpToDate, Savarese DMF (Ed), These results remain robust after both deterministic and UpToDate, Waltham, MA. 3. Corless CL, et al. Pathologic and molecular features correlate with long- probabilistic sensitivity analysis of various factors influ - term outcome after adjuvant therapy of resected primary GI stromal encing both cost and effectiveness of the competing strat - tumor: the ACOSOG Z9001 trial. J Clin Oncol. 2014;32(15):1563–70. egies. In aggregate, these results may be useful to guide 4. Blanke CD, et al. Phase III randomized, intergroup trial assessing imatinib mesylate at two dose levels in patients with unresectable or metastatic personalized clinical discussions between physicians and gastrointestinal stromal tumors expressing the kit receptor tyrosine patients considering potentially morbid surgery after a kinase: s0033. J Clin Oncol. 2008;26(4):626–32. period of neoadjuvant therapy in rectal GIST, especially 5. Joensuu H, et al. Adjuvant imatinib for high-risk GI stromal tumor: analysis of a randomized trial. J Clin Oncol. 2016;34(3):244–50. in patients with a strong aversion to surgical morbidity 6. Patel S. Long-term efficacy of imatinib for treatment of metastatic GIST. from abdominoperineal resection. Cancer Chemother Pharmacol. 2013;72(2):277–86. 7. Miller AR, et al. Quality of life and cost effectiveness analysis of therapy for locally recurrent rectal cancer. Dis Colon Rectum. 2000;43(12):1695–701. Abbreviations 8. Rutkowski P, et al. Neoadjuvant imatinib in locally advanced gastrointesti- GIST: Gastrointestinal stroma tumor; SGD: Singapore dollars; UAPR: Upfront nal stromal tumors (GIST ): the EORTC STBSG experience. Ann Surg Oncol. abdominoperineal resection; CIUP: Continued imatinib until progression; 2013;20(9):2937–43. QALY: Quality adjusted life years; WTP: Willingness to pay; PFS: Progression free 9. Wang D, et al. Phase II trial of neoadjuvant/adjuvant imatinib mesylate survival; OS: Overall survival; ICER: Incremental cost effectiveness ratio; NCCS: for advanced primary and metastatic/recurrent operable gastrointestinal National Cancer Centre Singapore; CT: Computed tomography; DSA: Deter- stromal tumors: long-term follow-up results of Radiation Therapy Oncol- ministic sensitivity analysis; PSA: Probabilistic sensitivity analysis. ogy Group 0132. Ann Surg Oncol. 2012;19(4):1074–80. Acknowledgements Nil. F arid et al. Clin Sarcoma Res (2020) 10:13 Page 11 of 11 10. Morgan, J.R., CP. Adjuvant and neoadjuvant imatinib for gastrointestinal 19. Chabot I, LeLorier J, Blackstein ME. The challenge of conducting stromal tumors. UpToDate, Savarese DMF (Ed), UpToDate, Waltham, MA. pharmacoeconomic evaluations in oncology using crossover trials: the 11. Gastrointestinal Stromal Tumor Meta-Analysis Group. Comparison of two example of sunitinib for gastrointestinal stromal tumour. Eur J Cancer. doses of imatinib for the treatment of unresectable or metastatic gastro- 2008;44(7):972–7. intestinal stromal tumors: a meta-analysis of 1,640 patients. J Clin Oncol. 20. Richardson SJ, et al. Activity of thalidomide and lenalidomide in mantle 2010;28(7):1247–53. cell lymphoma. Acta Haematol. 2010;123(1):21–9. 12. Majer IM, et al. Cost-effectiveness of 3-year vs 1-year adjuvant therapy 21. Briggs AH, et al. Model parameter estimation and uncertainty analysis: a with imatinib in patients with high risk of gastrointestinal stromal tumour report of the ISPOR-SMDM Modeling Good Research Practices Task Force recurrence in the Netherlands; a modelling study alongside the SSGXVIII/ Working Group-6. Med Decis Making. 2012;32(5):722–32. AIO trial. J Med Econ. 2013;16(9):1106–19. 22. Wyld L, Audisio RA, Poston GJ. The evolution of cancer surgery and future 13. Huse DM, et al. Cost effectiveness of imatinib mesylate in the treat - perspectives. Nat Rev Clin Oncol. 2015;12(2):115–24. ment of advanced gastrointestinal stromal tumours. Clin Drug Investig. 23. Huynh TK, et al. Primary localized rectal/pararectal gastrointestinal stro- 2007;27(2):85–93. mal tumors: results of surgical and multimodal therapy from the French 14. Sonnenberg FA, Beck JR. Markov models in medical decision making: a Sarcoma group. BMC Cancer. 2014;14:156. practical guide. Med Decis Making. 1993;13(4):322–38. 24. Cavnar MJ, et al. Rectal Gastrointestinal Stromal Tumor (GIST ) in the Era 15. Demetri GD, et al. Efficacy and safety of sunitinib in patients with of Imatinib: organ Preservation and Improved Oncologic Outcome. Ann advanced gastrointestinal stromal tumour after failure of imatinib: a Surg Oncol. 2017;24(13):3972–80. randomised controlled trial. Lancet. 2006;368(9544):1329–38. 25. Thanopoulou E, Judson I. The safety profile of imatinib in CML and GIST: 16. Demetri GD, et al. Efficacy and safety of regorafenib for advanced gastro - long-term considerations. Arch Toxicol. 2012;86(1):1–12. intestinal stromal tumours after failure of imatinib and sunitinib (GRID): 26. Garrison LP Jr, et al. An Overview of Value, Perspective, and Decision an international, multicentre, randomised, placebo-controlled, phase 3 Context-A Health Economics Approach: an ISPOR Special Task Force trial. Lancet. 2013;381(9863):295–302. Report [2]. Value Health. 2018;21(2):124–30. 17. Neumann PJ, Cohen JT, Weinstein MC. Updating cost-effectiveness–the curious resilience of the $50,000-per-QALY threshold. N Engl J Med. Publisher’s Note 2014;371(9):796–7. Springer Nature remains neutral with regard to jurisdictional claims in pub- 18. Wilson J, et al. Imatinib for the treatment of patients with unresectable lished maps and institutional affiliations. and/or metastatic gastrointestinal stromal tumours: systematic review and economic evaluation. Health Technol Assess. 2005;9(25):1–142. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Clinical Sarcoma Research Springer Journals

Treatment of gastrointestinal tumor (GIST) of the rectum requiring abdominoperineal resection following neoadjuvant imatinib: a cost-effectiveness analysis

Loading next page...
 
/lp/springer-journals/treatment-of-gastrointestinal-tumor-gist-of-the-rectum-requiring-V1UfYZ5P0G
Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2020
eISSN
2045-3329
DOI
10.1186/s13569-020-00135-7
Publisher site
See Article on Publisher Site

Abstract

Background: Neoadjuvant imatinib for gastrointestinal stromal tumors (GIST ) of the rectum can reduce, but may not eliminate, risk of surgical morbidity from permanent bowel diversion. We sought to evaluate the cost-effectiveness of alternative strategies in rectal GIST patients requiring abdominoperineal resection following neoadjuvant imatinib. Methods: We developed a Markov model using a healthcare payers’ perspective to estimate costs in 2017 Singapore dollars (SGD) and quality adjusted life years (QALYs) for upfront abdominoperineal resection (UAPR) versus continued imatinib until progression (CIUP) following 1 year of neoadjuvant imatinib. Transition probabilities and utilities were obtained from published data, and costs were estimated using data from the National Cancer Centre Singapore. Deterministic and probabilistic sensitivity analyses were conducted to probe model uncertainty. Incremental cost- effectiveness ratio below SGD 50,000 per QALY gained was considered cost-effective. Results: In the base case, UAPR dominates CIUP being both more effective (8.66 QALYS vs 5.43 QALYs) and less expensive (SGD 312,627 vs SGD 339,011). These estimates were most sensitive to 2 variables, utility of abdominoper- ineal resection and annual recurrence probability post-abdominoperineal resection. However, simultaneously varying the values of these variables to maximally favor CIUP did not render it the more cost effective strategy at willingness to pay ( WTP) of SGD 50,000. In probabilistic sensitivity analysis, UAPR had probability of being cost-effective com- pared with CIUP greater than 95%, reaching 100% at WTP SGD 10,000. Conclusion: UAPR is more effective and less costly than CIUP for patients with rectal GIST requiring abdominoper - ineal resection following neoadjuvant imatinib, and is the strategy of choice in this setting. Keywords: Rectal GIST treatment, Imatinib, Cost effectiveness analysis Background million per year [1]. The vast majority of cases arise in Gastrointestinal stromal tumors (GIST) are mesenchy- the stomach (40–60%), or jejunum/ileum (25–30%); mal tumors arising from the interstitial cells of Cajal approximately 5–15% of cases arise from the colon or in the alimentary tract with an incidence of 10–15 per rectum [2]. The primary treatment of GIST is com - plete surgical resection; among individuals without high-risk clinicopathologic features for recurrence this *Correspondence: Mohamad.farid@singhealth.com.sg approach results in sustained disease control in up to Division of Medical Oncology, National Cancer Centre Singapore, 11 70% of cases [3]. Imatinib, the tyrosine kinase inhibi- Hospital Drive, Singapore 169610, Singapore Full list of author information is available at the end of the article tor antagonizing the KIT and PDGFR oncoproteins © The Author(s) 2020. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/publi cdoma in/ zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Farid et al. Clin Sarcoma Res (2020) 10:13 Page 2 of 11 implicated in the majority of GISTs, has revolution- would require abdominoperineal resection for surgical ized the treatment of this previously dismal disease. In extirpation. advanced disease, imatinib achieves objective response rates, progression free survival (PFS) and overall sur- vival (OS) of 50%, 1.5 years and 4 years, respectively [4]. Methods In addition, 3 year of adjuvant imatinib improves over- Theoretical model/model overview all survival in patients with high-risk localized disease A comprehensive Markov model was developed to ana- who have undergone resection [5]. The achievement of lyze the health and cost impact of alternative strategies this significant therapeutic advance has come with the in management of rectal GIST from a healthcare sec- convenience of a pill taken once daily having an excel- tors’ perspective in Singapore. The structure of a Markov lent therapeutic index and minimal cumulative toxicity model is composed of mutually exclusive health states [6]. GISTs arising from the rectum poses a therapeutic relevant to the impact of disease and treatments under challenge in view of potential for significant surgical evaluation. Over a pre-determined time horizon, any morbidity from loss of bowel-sphincteric control and patient can stay in only 1 health state at any time, but may need for permanent bowel diversion if abdominoper- move between health states over time [14]. We modelled ineal resection is required for complete resection [7]. a hypothetical cohort of 60-year old patients with rectal Neoadjuvant imatinib has been shown to be safe and GIST who had received 1  year of neoadjuvant imatinib efficacious in the treatment of GIST presenting chal - at 400 milligrams daily with no disease progression, for lenges to upfront surgery, including disease arising which resection would necessitate an abdominoperineal from the rectum [8, 9]. Expert guidelines suggest that resection. The model evaluated 2 treatment options - neoadjuvant imatinib be given for up to 12  months continued imatinib until progression (CIUP), or surgi- in such cases to maximize response prior to surgery, cal resection with upfront abdominoperineal resection with consideration given for more imatinib postopera- (UAPR). (Figure 1) The model had a cycle length of 1 year tively to achieve a total of 3  years of imatinib therapy and followed patients over a time horizon of 20 years. A [10]. Although surgical resection remains the corner- total of 12 health states were defined: UAPR at 1st year stone of curative therapy, some patients continue to (“UAPR_Yr1”), UAPR at 2nd year (“UAPR_Yr2”) and defer curative surgery at the end of this period in view UAPR at 3rd year and beyond (“UAPR_Yr3+”) repre- of the excellent therapeutic index of imatinib and fear sent patients at different time-points following upfront of permanent morbidity from APR, the need for which abdominoperineal resection. The fourth health state may persist even after maximal tumor response to neo- accounts for patients receiving salvage surgery for their adjuvant imatinib. These patients may elect to con - 1st local recurrence following upfront abdominoperineal tinue imatinib indefinitely in lieu of upfront surgery, to resection (“Salvage Sx following LR1 post UAPR_Yr3+”). avoid surgical morbidity unless absolutely necessary, if The fifth health state addresses patients receiving the and when they progress on imatinib. Notwithstanding second strategy, CIUP (“CIUP”). The sixth and seventh the excellent therapeutic index of imatinib, there are health states contains patients following local progression no data to suggest that indefinite imatinib can substi - on CIUP–either undergoing abdominoperineal resection tute surgery as curative treatment in localized disease. following local progression on CIUP (“abdominoperineal The natural history for patients receiving imatinib in resection following LR on CIUP”) or subsequently sal- advanced disease is the eventual development of resist- vage surgery following 1st local recurrence after abdomi- ance and consequent local as well as distant progres- noperineal resection on CIUP (“Salvage Sx following LR1 sion. The development of metastatic disease precludes post CIUP”). The eighth through eleventh health states therapy with curative intent and leads to eventual encompasses patients at different lines of palliative ther - morbidity and mortality [11]. While imatinib has been apy in advanced disease, namely: distant recurrence (“1st shown to be cost-effective in both the adjuvant [12] and DR”) where they receive imatinib 400 milligrams daily advanced [13] disease setting, there are no data on the [11]; 1st progression in metastatic disease (“mets disease cost-effectiveness of imatinib as primary therapy in a 1st PD”) where they receive imatinib 800 milligrams daily setting where upfront surgery has significant potential [11]; 2nd progression in metastatic disease (“mets disease morbidity, specifically in rectal GIST. 2nd PD”) where they receive sunitinib 50 milligrams daily We sought to estimate, using a Markov decision 4 weeks on and 2 weeks off in a 6-weekly cycle [15]; and model, the relative effectiveness and cost-effectiveness 3rd progression in metastatic disease (“mets disease 3rd of upfront abdominoperineal resection compared with PD”) where they receive regorafenib 160 milligrams daily continued imatinib until progression for patients with 3  weeks on and 1  week off in a 4-weekly cycle [16]. The rectal GIST having received neoadjuvant imatinib who twelfth and final health state represents death (“Dead”). F arid et al. Clin Sarcoma Res (2020) 10:13 Page 3 of 11 Fig. 1 Markov model structure evaluating 2 treatment strategies (UAPR and CIUP) with 12 health states Farid et al. Clin Sarcoma Res (2020) 10:13 Page 4 of 11 For patients on UAPR, they received 2 years of postop- the former study much more closely resembled that used erative imatinib, to make up a total of 3 years of neoadju- in our model (12  months of imatinib at 400 milligrams vant and adjuvant imatinib, as consistent with treatment daily), as opposed to the 2 months of 600 milligrams daily guidelines [10]. Patients with disease progression while neoadjuvant imatinib employed in the RTOG 0132 study on adjuvant imatinib have a dismal prognosis, and thus [9]. In addition, there were 33 patients with rectal GIST transit directly to the health state where they receive in the retrospective study, as opposed to only 3 with rec- escalated doses of imatinib, “mets disease 1st PD”, regard- tal disease in RTOG 0132. For these reasons, we decided less of whether the progression was local or distant. The to use data obtained from the retrospective study in our first local recurrence after patients complete 2  years of model. The range of values considered for these transi - adjuvant imatinib therapy can yet be treated with salvage tion probabilities were derived from the confidence inter - surgery followed by 3 further years of adjuvant imatinib, vals reported in the literature or estimated if published a strategy with demonstrated benefit in newly diagnosed data were unavailable (Table 1). high-risk resected GIST [5]. The second local recurrence The model simulated transitions between health states or distant recurrence would lead to patients entering the with a cycle length of 1 year, which was chosen as a clini- metastatic disease health state receiving first line pallia - cally and therapeutically relevant interval for evaluation, tive therapy with imatinib, ‘1st DR’. considering the duration of GIST neoadjuvant and adju- For patients on CIUP, local progression would be vant (1 and 2  years, respectively) treatment and average treated with abdominoperineal resection followed by duration of disease control. The probability of transition - no further Imatinib. The first local recurrence following ing to death during each cycle was defined as the maxi - such an abdominoperineal resection would be treated mum value of observed mortality rate by using survival with salvage surgery only. The second local recurrence or data from studies in GIST as well as the background distant recurrence would lead to patients directly enter- mortality rate, estimated from age-specific death rates in ing the advanced disease health state receiving esca- Singapore [20]. The mortality of abdominoperineal resec - lated doses of imatinib (“mets disease 1st PD”), as these tion and subsequent salvage surgery for first post-abdom - patients have demonstrated failure to treatment with inoperineal resection local recurrence was assumed to be imatinib 400 milligrams daily. Tunnels states were used 0.5%. to facilitate the undertaking of salvage surgery in the set- ting of local recurrence (following APR in both the APR Utilities and costs and IUP arms) only once. QALYs were calculated by multiplying the time spent in a Model outcomes were defined as treatment costs (in given state (in life-years) by the utility score (a health sta- 2017 SGD) and quality-adjusted life years (QALYs). tus value ranging from 0 for death to 1 for perfect health) We calculated the incremental cost-effectiveness ratios associated with that state. The utility weights of all health (ICERs) of the more expensive to the less expensive states were derived from published studies (Table  1). strategy as the difference in costs divided by the differ - Utility associated with abdominoperineal resection was ence in QALYs – to compare 2 treatment strategies. All obtained from studies performed in recurrent rectal can- costs and health outcomes were discounted by 3% annu- cer [7]. Costs of surgery were estimated based on data ally. We adopted a willingness to pay (WTP) and ICER of from the National Cancer Centre Singapore (NCCS). SGD 50,000 per QALY gained as the threshold for cost- Drug costs were derived from the NCCS pharmacy data. effectiveness [17]. The model was developed using the Clinical consultation, blood tests and computed tomog- decision analytic software TreeAge Pro 2017 (TreeAge raphy (CT) scan performed every 3 monthly as surveil- Software, Williamstown, MA). lance were estimated to cost 3000 SGD per year based on expert opinion and NCCS data. Model transitions and survival estimates Event rates for progression and mortality were used to calculate transition probabilities [18, 19]. For advanced Sensitivity analysis GIST, data was obtained from randomized controlled To evaluate the robustness of the model and address the trials in the first [4], second [15] and third [16] line set - uncertainty in estimation of variables, we performed ting for GIST treatment. Recurrence rates following neo- a series of one-way deterministic sensitivity analyses adjuvant imatinib and surgery for localized GIST were (DSA), in each instance varying the value of one param- obtained from a large retrospective study of neoadjuvant eter at a time over its defined range and examining the imatinib in GIST [8]. Although there has been a prospec- effect of each parameter individually on ICERs for all tive single arm phase 2 study of neoadjuvant imatinib in variables. Where available, the range of values consid- localized GIST [9], the perioperative imatinib regimen in ered for these sensitivity analyses were derived from F arid et al. Clin Sarcoma Res (2020) 10:13 Page 5 of 11 Table 1 Transition probabilities, utilities and costs Variable Base case Range Distribution Reference for sensitivity for probabilistic analysis sensitivity analysis Probabilities Annual probability of recurrence post abdominoperineal resection 0.0871 0.0435–0.130 Beta Rutkowski et al. [8] Annual conditional probability of local recurrence post abdominoper- 0.135 0.0675–0.203 Uniform Rutkowski et al. [8] ineal resection Annual probability of 1st progression in metastatic GIST 0.370 0.296–0.444 Beta Blanke et al. [4] Annual probability of 2nd progression in metastatic GIST 0.811 0.649–0.973 Beta Blanke et al. [4] Annual probability of 3rd progression in metastatic GIST 0.708 0.566–0.850 Beta Demetri et al. [13] Annual probability of death in metastatic GIST post-regorafenib 0.405 0.270–0.410 Beta Demetri et al. [14] Utilities Recurrence-free health state post abdominoperineal resection 0.830 0.650–1 Beta Miller et al. [7] Recurrence-free health state on continued imatinib until progression 0.935 0.750–1 Beta Wilson et al. [17] GIST recurrence 0.748 0.598–0.898 Beta Majer et al. [10] GIST 1st progression in metastatic disease 0.712 0.685–0.739 Beta Chabot et al. [18] GIST 2nd progression in metastatic disease 0.712 0.685–0.739 Beta Assumption GIST 3rd progression in metastatic disease 0.712 0.685–0.739 Beta Assumption COSTS (SGD) Annual cost of imatinib 400 mg once daily 37 040 7 408–44 448 Gamma NCCS data Annual cost of sunitinib 50 mg once daily 4 weeks on, 2 weeks off 64 063 51 250–76 876 Gamma NCCS data Annual cost of regorafenib 160 mg once daily 3 weeks on, 1 week off 72 001 57 601–86 401 Gamma NCCS data Abdominoperineal resection 38 000 30 400 –45 600 Gamma NCCS data Salvage surgery (following 1st local recurrence post abdominoperineal 38 000 30 400–45 600 Gamma NCCS data resection) Annual cost of follow-up (consultation, blood tests, computer tomog- 3 000 2 400–3 600 Gamma NCCS data raphy scans every 3 months) mg milligrams, GIST gastrointestinal stromal tumor, NCCS National Cancer Centre Singapore, SGD Singapore dollars ± 50% (wider intervals used in estimation due to paucity of systematic data for recurrences specifically in rectal GIST post neoadjuvant imatinib) ± 20% + 20% and −80% (lower bound for imatinib extended to −80% to account for possible significant decrease in imatinib cost with advent of generic imatinib) the confidence intervals reported in the primary litera - Results ture (Table 1). Base case/main analysis To account for variation in multiple parameters Examining the 2 treatment strategies over the 20-year simultaneously, we also completed probabilistic sen- time horizon, UAPR (SGD 312,627) is less costly com- sitivity analyses (PSA). We performed 10,000 Monte pared to CIUP (SGD 339,011), yet generates more QALYs Carlo simulations, each time randomly sampling from (8.66 vs 5.43  years), thus dominating over CIUP as a the distributions for all parameters simultaneously. strategy. Uncertainty in utilities was represented by β distribu- tions which are bounded by 0 and 1, while uncertainty in cost was represented by γ distributions, which are Sensitivity analysis bounded by 0 and infinity [21]. To succinctly represent One-way DSA revealed our results to be robust for all uncertainty surrounding our base case estimate of cost- variables across the pre-defined range of values (Table  1), effectiveness, cost-effectiveness acceptability curves with UAPR consistently dominating over CIUP; the only were derived and used to project the probability that exception to this arose in the context of annual recur- each treatment strategy was economically preferred rence probability post abdominoperineal resection under various WTP thresholds. greater than 12.5%, where an ICER for UAPR emerged, Farid et al. Clin Sarcoma Res (2020) 10:13 Page 6 of 11 Fig. 2 One way sensitivity analysis evaluating incremental cost effectiveness ratios (ICERs) of UAPR compared with CIUP for a range of utilities associated with abdominoperineal resection though remaining well below SGD 50,000 per QALY The probability of UAPR being cost-effective compared gained. with CIUP is 100% for WTP SGD 10,000 and above, and We performed several scenario analyses to further never drops below 95%. Consistent with the findings delineate the impact of extreme values for several vari- from the base case analysis, the cost-effectiveness plane ables on our outcome. When considering extremely low reveals UAPR to be dominant over CIUP for the majority values of utility for the abdominoperineal resection, of iterations of the PSA at a WTP of SGD 50,000. CIUP generated more QALYs than UAPR for utility val- ues below 0.34, with CIUP producing ICER less than SGD Discussion 50,000 for utility values less than 0.24 (Fig. 2 ). As earlier In this cost-effectiveness analysis, considering rec - described, UAPR ceased to dominate CIUP when post- tal GIST patients with no disease progression follow- abdominoperineal resection recurrence exceeded 12%. ing 12  months of neoadjuvant imatinib who required Extending this analysis to consider even higher annual abdominoperineal resection for surgical extirpation, probabilities of recurrence post abdominoperineal resec- UAPR is more effective and less costly than CIUP. These tion, UAPR generated more QALYs for annual recurrence results remained robust after considering one-way DSA probabilities below 47.0%; for higher annual recurrence of multiple factors. The only exception to this arose when probabilities, CIUP became the dominant strategy. UAPR the annual probability of recurrence following abdomi- remained more cost-effective than CIUP at WTP of SGD noperineal resection exceeded 12.5%; in this situation, 50,000 for annual recurrence probabilities below 31% UAPR was costlier and more effective than CIUP, with (Fig.  3). Additionally, we performed a 2-way sensitivity ICER values remaining well below the SGD 50,000 WTP analysis exploring extreme values of utility for abdomin- threshold. These findings were replicated in PSA, with operineal resection (as low as 0.5) and annual recurrence the probability of UAPR being cost-effective compared probabilities post abdominoperineal resection (as high as with CIUP being 100% from a WTP threshold of SGD 35%). In this analysis, the net monetary benefit for UAPR 10,000 onwards. Taken in aggregate, our data suggest would always be superior for annual recurrence probabil- that, consistent with current practice, surgical extirpation ities lower than 18% (Fig. 4). must necessarily remain the primary curative treatment From the PSA, the cost-effectiveness acceptability modality in localized rectal GIST. Such a conclusion curve is shown in Fig. 5. would be consistent with the practice and evidence from the treatment of solid tumors in general [22], and GIST F arid et al. Clin Sarcoma Res (2020) 10:13 Page 7 of 11 Fig. 3 One way sensitivity analysis evaluating incremental cost effectiveness ratios (ICERs) of UAPR compared with CIUP for a range of annual recurrence probabilities post abdominoperineal resection. Below a recurrence probability of 12%, UAPR dominates CIUP–it costs less and is more effective, thus generating no meaningful ICER. The range of recurrence probabilities considered thus begins with 15% in particular [3]. Complete resection should thus not be even if a 50% uncertainty of this point estimate is con- forsaken in the treatment of rectal GIST. sidered as an estimate of a 95% confidence interval, the To account for the possibility of widespread individ- upper bound of this confidence interval would be 13.0%. ual variation in utility of abdominoperineal resection, Considering other data, the annual probability of recur- we performed one way scenario analyses to explore the rence for high risk GIST resected upfront followed by impact of extreme aversion to abdominoperineal resec- 3 years of adjuvant imatinib in the landmark prospective tion through evaluation of very low values of utility asso- study of adjuvant imatinib was 6.9% [5], and the annual ciated with abdominoperineal resection. We found that recurrence probability of recurrence amongst imatinib- CIUP would be more cost effective than UAPR (WTP of treated rectal GIST patients in another retrospective SGD 50,000) only if utility values were 0.24 or lower. This study was 5.7% [23]. These data suggest it extremely hypothetical value, assigning extraordinarily severe disu- improbably that annual recurrence rates post resection of tility to abdominoperineal resection, is much lower than rectal GIST treated with imatinib would be high enough the value of 0.83 described in the literature [7], and may for UAPR to be rendered cost-ineffective compared with be argued to be implausible in clinical practice. Nonethe- CIUP. less, we think that this information may be useful as a Probing the sensitivity to these variables further, we quantitative threshold to guide individualized discussions simultaneously explored extreme values of both util- between physicians and patients with extreme aversion ity associated with abdominoperineal resection and to abdominoperineal resection following neoadjuvant annual recurrence probabilities post abdominoperineal imatinib for rectal GIST. resection in a 2-way sensitivity analysis. In this analysis, As earlier described, UAPR ceased to dominate CIUP UAPR was shown to be the optimal strategy at a WTP for annual recurrence probabilities following abdomi- of SGD 50,000 for all cases when post abdominoperineal noperineal resection greater than 12.5%. When more resection recurrence was less than 18% annually, regard- extreme values of recurrence probabilities were evaluated less of how low the utility associated with abdominop- in a scenario analysis, the ICER of UAPR remained below erineal resection was (considering utility values as low SGD 50,000 for recurrence probabilities lower than as 0.5). Even when considering the lower bound of the 31.5%. By comparison, the annual recurrence probability confidence interval for utility associated with abdomi - following resection after neoadjuvant imatinib is 8.7% [8]; noperineal resection (0.65) [7], UAPR would remain Farid et al. Clin Sarcoma Res (2020) 10:13 Page 8 of 11 Fig. 4 Two way sensitivity analysis comparing the net monetary benefit (NMB) of UAPR vs CIUP at a willingness to pay of SGD 50,000 when simultaneously considering varying values of utility associated with abdominoperineal resection and annual recurrence probabilities post abdominoperineal resection. Red denotes UAPR having superior NMB, while blue denotes CIUP having superior NMB more cost-effective for annual recurrence probabili - (Fig.  6). As ICER is calculated by dividing the difference ties post abdominoperineal resection below 24%. Either in costs by the difference in effectiveness, and given that way, these annual recurrence probabilities below which CIUP was less effective than UAPR, this suggests that UAPR would be the optimal strategy are well beyond the the cost of the CIUP strategy was greater than the UAPR upper limit of the confidence interval for annual recur - strategy, even if imatinib were free. This increased cost is rence probability of 13%, reinforcing the robustness likely related to the costs accruing from inferior clinical of the result favoring the superiority of UAPR. In addi- outcomes with CIUP. tion, notwithstanding the improbability of such high There are several limitations to our study. Our analy - annual recurrence rates in clinical practice, these data sis applies only to patients for whom abdominoperineal would be useful in therapeutic discussions with clinico- resection is required following neoadjuvant imatinib, a pathologically high risk rectal GIST patients expressing situation that applies to the minority of patients. In the severe aversion to the morbidity associated with abdomi- imatinib era, with the excellent clinical responses and noperineal resection. Formally quantifying individual long term disease control afforded by imatinib, there preferences, and comparing it against quantitated bound- are a range of less morbid surgical procedures avail- aries of transition probabilities associated with particular able to patients without compromising completeness of strategies, will hopefully enable more informed and pre- surgical extirpation. In one of the largest series of rec- cise therapeutic decision making. tal GIST reported, the rate of patients requiring APR The cost of imatinib was not a significant factor in or pelvic exenteration in the imatinib era was only 3%, determining the optimal strategy in this analysis. Since compared with 59% before the advent of imatinib [24]. CIUP was less effective than UAPR, CIUP would not be Nonetheless, our aim was to very specifically evaluate more cost effective even if it was the less costly strat - the outcomes for the most morbid surgical scenario— egy. As it were, when cost effectiveness of CIUP rela - loss of bowel continence—to see if deferring surgery tive to UAPR was evaluated in relation to how the cost could possibly be the more cost-effective option fol - of imatinib varied in a one-way sensitivity analysis, there lowing neoadjuvant imatinib, an option that patients was no faced with such a clinical scenario may be inclined to meaningful ICER accrued (the ICER remained nega- consider (as distinct from the situation when surgery tive) even when the cost of imatinib was zero dollars has less morbidity, and thus opting for upfront surgery F arid et al. Clin Sarcoma Res (2020) 10:13 Page 9 of 11 Fig. 5 Cost effectiveness acceptability curve from probabilistic sensitivity analysis comparing UAPR and CIUP. UAPR has a 100% probability of being more cost effective that CIUP for willingness to pay of SGD 10,000 and above Fig. 6 One way sensitivity analysis evaluating incremental cost effectiveness ratios (ICERs) of CIUP compared with UAPR for a range of costs of imatinib. For all values of imatinib price, no meaningful ICER is accrued. In the setting of the known finding of CIUP having less effectiveness, this suggests that it is the more costly strategy overall regardless of imatinib cost is more clear-cut). As it turns out, our analysis confirms therapy, primarily because the toxicity from imatinib is the importance of complete surgical extirpation even if generally limited and tolerable [25]. Additionally, drug surgical morbidity is a certainty. We did not consider toxicities from sunitinib and regorafenib would only the utilities and costs of toxicity accruing from drug become relevant in the setting of advanced disease, and Farid et al. Clin Sarcoma Res (2020) 10:13 Page 10 of 11 Authors’ contributions would not differentially affect costs and effectiveness MF and DM conceptualized and designed the study. MF, JO, CC, GT, MT, across the 2 strategies. We did not consider indirect RQ and JT collected and assembled the data. MF and DM analyzed and costs such as loss of earnings from disrupted employ- interpreted the data. All authors wrote and approved the final manuscript. All authors read and approved the final manuscript. ment or costs to unpaid caregivers, given that we employed a healthcare sectors’ perspective as opposed Funding to a societal perspective in our analysis [26]. We did Nil, there was no funding support. not include disease related prognostic factors like size, Availability of data and materials mitotic rate and mutational status in our model. This is All data generated and/or analyzed during this study are included in this because full clinicopathologic characterization (patho- published article. logical size as well as mitotic count) is not available at Ethics approval and consent to participate the start of neoadjuvant imatinib for rectal GIST, where Not applicable diagnosis is often based on endoscopically acquired Consent for publication fine needle biopsy —so these factors do not dictate ini - Not applicable. tiation of imatinib in clinical practice. In addition, the primary value of mutational status is in ensuring only Competing interests RQ has received honoraria from, served in a consulting or advisory role for, patients with imatinib sensitive mutations (KIT muta- and been part of a speakers’ bureau for Novartis, Bayer, Merck and Eisai. RQ has tions, and PDGFRA mutations other than D842V) received honoraria and served in a consulting or advisory role for Bristol-Myers receive imatinib–this group forms the vast majority Squibb (BMS). RQ has received travel, accommodation and expenses from Roche, Novartis and Eisai. RQ has received education grants from Novartis, (> 80%) of GIST patients. There were, however, patient Eisai, Janssen and Bayer. (surgical fitness, specific anatomical factors, comorbid - ities and resources for self-care, perceptions of surgical Author details Division of Medical Oncology, National Cancer Centre Singapore, 11 Hospital morbidity) and treatment (surgical expertise) prognos- Drive, Singapore 169610, Singapore. Division of Surgical Oncology, National tic factors that we did not account for. It may be argued Cancer Centre, Singapore, Singapore. Parkway Cancer Centre, Singapore, that a WTP of SGD 50,000 is too conservative for con- Singapore. Radiation Oncology, Farrer Park Hospital, Singapore, Singapore. Programme in Health Services and Systems Research, Graduate Medical temporary considerations of cost-effectiveness [17]; School, Duke-National University of Singapore, Singapore, Singapore. however, given the robustness of our results, the con- sideration of higher levels of WTP is not likely to alter Received: 13 February 2018 Accepted: 30 July 2020 our conclusions. Conclusions References In summary, our cost-effectiveness analysis confirms the 1. Soreide K, et al. Global epidemiology of gastrointestinal stromal tumours importance of surgical extirpation in rectal GIST patients (GIST ): a systematic review of population-based cohort studies. Cancer who have received neoadjuvant imatinib, showing Epidemiol. 2016;40:39–46. 2. Morgan, J., Raut CP. Epidemiology, classification, clinical presenta- upfront abdominoperineal resection to be more effective tion, prognostic features, and diagnostic work-up of gastrointestinal and less costly than continued imatinib until progression. mesenchymal neoplasms including GIST. UpToDate, Savarese DMF (Ed), These results remain robust after both deterministic and UpToDate, Waltham, MA. 3. Corless CL, et al. Pathologic and molecular features correlate with long- probabilistic sensitivity analysis of various factors influ - term outcome after adjuvant therapy of resected primary GI stromal encing both cost and effectiveness of the competing strat - tumor: the ACOSOG Z9001 trial. J Clin Oncol. 2014;32(15):1563–70. egies. In aggregate, these results may be useful to guide 4. Blanke CD, et al. Phase III randomized, intergroup trial assessing imatinib mesylate at two dose levels in patients with unresectable or metastatic personalized clinical discussions between physicians and gastrointestinal stromal tumors expressing the kit receptor tyrosine patients considering potentially morbid surgery after a kinase: s0033. J Clin Oncol. 2008;26(4):626–32. period of neoadjuvant therapy in rectal GIST, especially 5. Joensuu H, et al. Adjuvant imatinib for high-risk GI stromal tumor: analysis of a randomized trial. J Clin Oncol. 2016;34(3):244–50. in patients with a strong aversion to surgical morbidity 6. Patel S. Long-term efficacy of imatinib for treatment of metastatic GIST. from abdominoperineal resection. Cancer Chemother Pharmacol. 2013;72(2):277–86. 7. Miller AR, et al. Quality of life and cost effectiveness analysis of therapy for locally recurrent rectal cancer. Dis Colon Rectum. 2000;43(12):1695–701. Abbreviations 8. Rutkowski P, et al. Neoadjuvant imatinib in locally advanced gastrointesti- GIST: Gastrointestinal stroma tumor; SGD: Singapore dollars; UAPR: Upfront nal stromal tumors (GIST ): the EORTC STBSG experience. Ann Surg Oncol. abdominoperineal resection; CIUP: Continued imatinib until progression; 2013;20(9):2937–43. QALY: Quality adjusted life years; WTP: Willingness to pay; PFS: Progression free 9. Wang D, et al. Phase II trial of neoadjuvant/adjuvant imatinib mesylate survival; OS: Overall survival; ICER: Incremental cost effectiveness ratio; NCCS: for advanced primary and metastatic/recurrent operable gastrointestinal National Cancer Centre Singapore; CT: Computed tomography; DSA: Deter- stromal tumors: long-term follow-up results of Radiation Therapy Oncol- ministic sensitivity analysis; PSA: Probabilistic sensitivity analysis. ogy Group 0132. Ann Surg Oncol. 2012;19(4):1074–80. Acknowledgements Nil. F arid et al. Clin Sarcoma Res (2020) 10:13 Page 11 of 11 10. Morgan, J.R., CP. Adjuvant and neoadjuvant imatinib for gastrointestinal 19. Chabot I, LeLorier J, Blackstein ME. The challenge of conducting stromal tumors. UpToDate, Savarese DMF (Ed), UpToDate, Waltham, MA. pharmacoeconomic evaluations in oncology using crossover trials: the 11. Gastrointestinal Stromal Tumor Meta-Analysis Group. Comparison of two example of sunitinib for gastrointestinal stromal tumour. Eur J Cancer. doses of imatinib for the treatment of unresectable or metastatic gastro- 2008;44(7):972–7. intestinal stromal tumors: a meta-analysis of 1,640 patients. J Clin Oncol. 20. Richardson SJ, et al. Activity of thalidomide and lenalidomide in mantle 2010;28(7):1247–53. cell lymphoma. Acta Haematol. 2010;123(1):21–9. 12. Majer IM, et al. Cost-effectiveness of 3-year vs 1-year adjuvant therapy 21. Briggs AH, et al. Model parameter estimation and uncertainty analysis: a with imatinib in patients with high risk of gastrointestinal stromal tumour report of the ISPOR-SMDM Modeling Good Research Practices Task Force recurrence in the Netherlands; a modelling study alongside the SSGXVIII/ Working Group-6. Med Decis Making. 2012;32(5):722–32. AIO trial. J Med Econ. 2013;16(9):1106–19. 22. Wyld L, Audisio RA, Poston GJ. The evolution of cancer surgery and future 13. Huse DM, et al. Cost effectiveness of imatinib mesylate in the treat - perspectives. Nat Rev Clin Oncol. 2015;12(2):115–24. ment of advanced gastrointestinal stromal tumours. Clin Drug Investig. 23. Huynh TK, et al. Primary localized rectal/pararectal gastrointestinal stro- 2007;27(2):85–93. mal tumors: results of surgical and multimodal therapy from the French 14. Sonnenberg FA, Beck JR. Markov models in medical decision making: a Sarcoma group. BMC Cancer. 2014;14:156. practical guide. Med Decis Making. 1993;13(4):322–38. 24. Cavnar MJ, et al. Rectal Gastrointestinal Stromal Tumor (GIST ) in the Era 15. Demetri GD, et al. Efficacy and safety of sunitinib in patients with of Imatinib: organ Preservation and Improved Oncologic Outcome. Ann advanced gastrointestinal stromal tumour after failure of imatinib: a Surg Oncol. 2017;24(13):3972–80. randomised controlled trial. Lancet. 2006;368(9544):1329–38. 25. Thanopoulou E, Judson I. The safety profile of imatinib in CML and GIST: 16. Demetri GD, et al. Efficacy and safety of regorafenib for advanced gastro - long-term considerations. Arch Toxicol. 2012;86(1):1–12. intestinal stromal tumours after failure of imatinib and sunitinib (GRID): 26. Garrison LP Jr, et al. An Overview of Value, Perspective, and Decision an international, multicentre, randomised, placebo-controlled, phase 3 Context-A Health Economics Approach: an ISPOR Special Task Force trial. Lancet. 2013;381(9863):295–302. Report [2]. Value Health. 2018;21(2):124–30. 17. Neumann PJ, Cohen JT, Weinstein MC. Updating cost-effectiveness–the curious resilience of the $50,000-per-QALY threshold. N Engl J Med. Publisher’s Note 2014;371(9):796–7. Springer Nature remains neutral with regard to jurisdictional claims in pub- 18. Wilson J, et al. Imatinib for the treatment of patients with unresectable lished maps and institutional affiliations. and/or metastatic gastrointestinal stromal tumours: systematic review and economic evaluation. Health Technol Assess. 2005;9(25):1–142. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions

Journal

Clinical Sarcoma ResearchSpringer Journals

Published: Aug 6, 2020

There are no references for this article.