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

Learn More →

Multiplex quantitative analysis of cancer-associated fibroblasts and immunotherapy outcome in metastatic melanoma

Multiplex quantitative analysis of cancer-associated fibroblasts and immunotherapy outcome in... Background: The cancer-associated fibroblast (CAF) population is implicated in immune dysregulation. Here, we test the hypothesis that CAF profiles in pretreatment tumor specimens are associated with response to immune checkpoint blockade of programmed cell death 1 (PD-1). Methods: Pretreatment whole tissue sections from 117 melanoma patients treated with anti-PD-1 therapy were assessed by multiplex immunofluorescence to detect CAFs defined by Thy1, smooth muscle actin (SMA), and fibroblast activation protein (FAP). Two independent image analysis technologies were used: inForm software (PerkinElmer) to quantify cell counts, and AQUA™ to measure protein by quantitative immunofluorescence (QIF). CAF parameters by both methodologies were assessed for association with previously measured immune markers (CD3, CD4, CD8, CD20, CD68, PD-L1), best overall response, progression-free survival (PFS), and overall survival (OS). Results: CAF parameters, by cell counts or QIF, did not correlate with immune markers nor with best overall response. However, both Thy1 and FAP cell counts had significant positive associations with PFS (all P < 0.05) and OS (all P < 0.003). SMA cell counts showed negative associations with outcome in anti-PD-1 treated patients. Similar associations were not observed in a control cohort of historical melanoma patients predating immunotherapy. Instead, FAP was a negative prognostic biomarker (P = 0.01) in the absence of immunotherapy. Multivariable analyses revealed significant PFS and OS associations with the CAF parameters were independent of baseline variables. Conclusions: Pretreatment CAF profiles are associated with melanoma immunotherapy outcome. Multiplex CAF analysis has potential as an objective companion diagnostic in immuno-oncology. Keywords: Biomarkers, Cancer-associated fibroblasts, Fibroblast activation protein, Immunotherapy, Melanoma, PD-1 Introduction protein 4 (CTLA-4), which is targeted by ipilimumab [4]. Immune checkpoint blockade has become a new stand- Nevertheless, clinical benefit is limited to ~ 40% of meta- ard in melanoma immunotherapy and the overall sur- static melanoma patients treated with anti-PD-1 therapy, vival of patients with metastatic disease has improved which is compounded by the lack of approved predictive from ~ 9 months before 2011 to greater than 3 years strategies [1, 5]. Due to widespread use of PD-1 blockade [1–3]. The tumor-infiltrating lymphocyte (TIL) popu- and its recent introduction into the adjuvant setting [6], lation expresses immune checkpoints, programmed there is an increasing need for robust biomarkers to in- cell death 1 (PD-1), which is targeted by pembrolizumab form the practice of precision immuno-oncology [7]. and nivolumab; and cytotoxic T-lymphocyte associated The cancer-associated fibroblast (CAF) population en- gages in a complex and poorly understood interplay with tumor cells and immune cells, and are the predominant * Correspondence: david.rimm@yale.edu stromal cell type within the tumor microenvironment. Department of Pathology, Yale School of Medicine, New Haven, CT 06510, CAFs are characterized by expression of Thy1, with sub- USA Yale Cancer Center, Yale School of Medicine, New Haven, CT 06510, USA sets expressing smooth muscle actin (SMA) or fibroblast Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 2 of 10 activation protein (FAP) [8, 9]. Thy1 is a glycophosphati- only included pretreatment formalin-fixed, paraffin- dylinositol (GPI)-anchored cell surface protein that binds embedded (FFPE) specimens after review by a board- to integrins and may be involved in cell–cell adhesion certified pathologist. All specimens were collected from [10]. SMA is a major component of the contractile ap- the Yale Pathology archives. Clinicopathological data paratus that allows fibroblasts to produce contractile were collected from clinical records and pathology re- force [11]. FAP is a type II transmembrane serine prote- ports; the data cut-off date was September 1, 2017. Re- ase that cleaves collagen I as an endopeptidase and en- sponse Evaluation Criteria in Solid Tumors (RECIST) gages in post-translational modification of neuropeptide 1.1 were used to determine best overall response as Y as a dipeptidyl peptidase, which is the rare ability to complete response (CR), partial response (PR), stable hydrolyze the post-proline bond two residues from the disease (SD), or progressive disease (PD), and objective N-terminus of substrates [12]. FAP is weakly expressed response rate (ORR; CR/PR) and disease control rate or not detected in normal adult tissues but is upregu- (DCR; CR/PR/SD) [19]. A historical cohort of 194 mel- lated at sites of activated stroma in tumors and in anoma patients, collected prior to the advent of anti-PD- chronic inflammation [13]. Emerging preclinical evidence 1, was used as the control group. Cohort characteristics implicates CAFs in immune dysregulation and response are detailed in Table 1. Other characteristics of the anti- to immunotherapy [14–16]. However, CAFs represent a PD-1 treated cohort including the melanoma specimen, heterogeneous group and different CAF subsets may have time interval to anti-PD-1 therapy, and prior immune opposing functions. A more comprehensive understand- checkpoint blockade are shown in Additional file 1: ing of different CAF subsets as well as their impact on hu- Table S1. All patients provided written informed consent man immunotherapy outcome is needed. or waiver of consent. The study was approved by the Yale We hypothesized that pretreatment CAF profiles of Human Investigation Committee protocol #9505008219 patient tumors would be associated with immunotherapy and conducted in accordance with the Declaration of outcome. However, predictive biomarkers strictly require Helsinki. statistical evidence from a formal test for interaction in randomized placebo-controlled studies, which are no Multiplex immunofluorescence CAF panel longer ethically possible for melanoma. Therefore, we 5-plex immunofluorescence using isotype-specific anti- tested a control cohort of historical melanoma patients bodies was performed on FFPE whole tissue sections for predating immunotherapy instead to distinguish prog- simultaneous detection of markers as previously described nostic value and show a specific association between the [20]. The protocol is detailed in the Additional file 1. biomarker and treatment outcome. We describe this type of biomarker as “indicative”, a separate category Image analysis by two independent methods: cell counts from truly predictive biomarkers under existing statis- versus quantitative immunofluorescence tical definitions [17]. Briefly, indicative value is demon- Cell counts were determined by the pattern recognition strated when: [1] the hazard ratio is statistically software, inForm Tissue Finder (PerkinElmer, Waltham, significant in the treatment cohort and is not significant MA, USA), on multispectral images acquired using a in the control cohort; or [2] the hazard ratio is statisti- Vectra 3 system (PerkinElmer) as previously described cally significant in both the treatment and control co- [21]. Multispectral images were decomposed into their horts, but the respective 95% confidence intervals do not various components by spectral unmixing using a digital significantly overlap. The former characteristic is purely spectral library consisting of spectral profiles of each of indicative, and the latter is both prognostic and indica- the fluorophores. Automated tissue segmentation identi- tive [17]. fied tumor and stroma regions. Cell segmentation within Here, we assess the clinical significance of CAFs for these regions identified individual cells and respective the prediction of immunotherapy outcome in metastatic nuclei, cytoplasm, and membrane components using sig- melanoma. We hypothesize that the expression of these nal in the nucleus and membrane as internal and exter- candidate biomarkers, Thy1, SMA, and FAP, will classify nal cell borders, then cells were phenotyped for marker anti-PD-1 therapy treated patients into groups that expression. Cell counts for each melanoma case were benefit and those that do not. calculated in terms of the number of cells positive for the marker of interest as a percentage of the cell popula- Methods tion in which it was measured. Protein expression of the Patient cohort various markers was determined by the automated quan- The study cohort is a retrospective collection of 117 titative analysis (AQUA) method of QIF on fluorescence melanoma patients treated with anti-PD-1 therapy in the images acquired using a PM-2000 system (Navigate metastatic setting between 2011 and 2017 at Yale Cancer BioPharma, Carlsbad, CA, USA) as previously described Center. Uveal melanoma was excluded [18]. The analysis [22]. A total compartment, consisting of all cells, or a Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 3 of 10 Table 1 Clinicopathological characteristics of the melanoma cohort treated with anti-PD-1 therapy and the control melanoma cohort for CAF profiling Characteristic Anti-PD-1 patients, No. (%) Objective response Disease control rate Control patients, No. (%) rate (CR/PR), No. (%) (CR/PR/SD), No. (%) Overall 117 (100) 55 (47) 81 (69) 194 (100) Age (y) < 65 67 (57) 34 (51) 51 (76) 87 (45) ≥ 65 50 (43) 21 (42) 30 (60) 107 (55) Sex Male 70 (60) 35 (50) 48 (69) 110 (57) Female 47 (40) 20 (43) 33 (70) 84 (43) Treatment Pembrolizumab 41 (35) 20 (49) 30 (73) 0 Nivolumab 18 (15) 7 (39) 9 (50) 0 Ipilimumab plus nivolumab 58 (50) 28 (48) 42 (72) 0 Prior immune checkpoint blockade Yes 36 (31) 13 (36) 22 (61) 0 No 81 (69) 42 (52) 59 (73) 194 (100) Mutation status BRAF 39 (33) 19 (49) 27 (69) NA NRAS 18 (15) 8 (44) 11 (61) NA KIT 2 (2) 1 (50) 2 (100) NA None detected 58 (50) 27 (47) 41 (71) NA Stage at diagnosis I 24 (21) 14 (58) 19 (79) 77 (40) II 23 (20) 12 (52) 16 (70) 80 (41) III 38 (32) 16 (42) 24 (63) 30 (15) IV 20 (17) 6 (30) 13 (65) 3 (2) Not available 12 (10) 7 (58) 9 (75) 4 (2) Abbreviations: CAF cancer-associated fibroblast, CR complete response, NA not available, PR partial response, SD stable disease Thy1 compartment was generated by automated process- (PFS) and overall survival (OS) functions were computed ing and thresholding of the DAPI signal or Thy1 signal, and comparisons were made by the log-rank test. Multi- respectively. QIF scores were calculated by dividing the variable Cox proportional hazards models included age, summed pixel intensities for the marker of interest by the sex, mutation status, stage, treatment, and prior immune area of the compartment in which it was measured [22]. checkpoint blockade as covariates [24–27]. All statistical Overall QIF scores were derived for each melanoma case tests were two-sided and statistical significance was de- by averaging scores from each field of view. fined as P < 0.05. Statistical analysis was performed using GraphPad Prism 7 (GraphPad Software, La Jolla, CA, Statistical analysis USA) and JMP Pro 13 (SAS Institute, Cary, NC, USA). Statistical comparisons for cell count and QIF data were The sample size of 117 patients had at least 80% power made using unpaired t-test or analysis of variance at P = 0.05 to detect a difference in means of 0.52 stand- (ANOVA) followed by Tukey’s test for multiple compari- ard deviations in each CAF parameter for responders sons as appropriate. The Lausen and Schumacher (CR/PR) versus non-responders (SD/PD). method of maximally selected rank statistics, a powerful nonparametric method for assessing predictive power of Results a continuous variable for a dependent variable, was used Correlation between cell counts and quantitative to determine thresholds to objectively define low and immunofluorescence high statuses for the measured CAF parameters [23]. Tissue biomarkers can be quantified in situ by counting Kaplan–Meier estimates of progression-free survival positive cells for the biomarker or in terms of quantitative Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 4 of 10 protein expression levels per unit area. These are two in- for prognostic value in place of a placebo arm. Simi- dependent types of parameters and may be nonequivalent lar associations were not observed in the control pa- in clinical significance. The relationship between cell tients (Fig. 3band Table 3). Remarkably, FAP cell count counts and QIF was assessed by linear regression, which was a significant negative prognostic biomarker in the ab- revealed a positive correlation for Thy1 (R = 0.35), SMA sence of immunotherapy (P = 0.01) with an inverted haz- 2 2 (R = 0.36), and FAP (R = 0.62) (Additional file 1: ard ratio (HR = 0.57, 95% CI, 0.37–0.88) relative to that of Figure S1A). On the contrary, there was no correlation be- the anti-PD-1 patients (HR = 4.11, 95% CI, 2.05–9.14) tween different markers, which confirmed their independ- (Table 3). Multivariable analyses further revealed signifi- ence (Additional file 1:Figure S1B). cant survival associations with the CAF parameters, par- ticularly for FAP, independent of age, sex, mutation, stage, Immune markers and CAF parameters treatment, and prior immune checkpoint blockade Pretreatment whole tissue sections from 117 melanoma (Tables 2–3). The QIF data showed similar trends in patients treated with anti-PD-1 therapy underwent CAF relation to survival (Additional file 1:FigureS7and (Thy1, SMA, FAP) profiling by multiplex immunofluor- Additional file 1:TablesS2–S3). Survival analysis by escence (Fig. 1). The relationship between CAFs and in- treatment group generally showed similar trends despite filtration of immune cell populations or expression of the reduction in statistical power (Additional file 1: immune markers in melanoma was assessed by linear Tables S4–S5). regression with previous data [17]. There was no cor- relation between the CAF parameters and CD3, CD4, Discussion CD8, CD20, CD68, or PD-L1, which confirmed their In this study, we determine the clinical significance of independence of those immune markers (Fig. 2 and pretreatment CAF (Thy1, SMA, FAP) profiles according Additional file 1: Figure S2). to both in situ cell counts and QIF protein expression in relation to immunotherapy outcome in metastatic mel- Best overall response by RECIST and CAF parameters anoma. PFS and OS had positive associations with Thy1 The CAF parameters, by cell counts or QIF, were ana- and FAP cell counts, and negative associations with lyzed in relation to specimen-specific variables and SMA cell count, which were specific to anti-PD-1 tumor burden classifications defined by RECIST 1.1 treated patients. Significant PFS and OS associations [19]. There were no significant associations with sex or with the CAF parameters were independent of age, sex, mutation status of melanoma patients for the CAF pa- mutation, stage, treatment, and prior immune check- rameters by cell counts or QIF (all P > 0.05; Additional point blockade [24–27]. While the two quantitative file 1: Figure S3). Neither Thy1, SMA, nor FAP cell methods are independent, cell counts correlated with counts were associated with best overall response defined by QIF and revealed concordant associations with response RECIST (all P > 0.05; Additional file 1:FigureS4A). The cor- and survival outcome. responding QIF data (Additional file 1:FigureS4B)and fur- This study attempts to rigorously investigate multiplex ther analyses on ORR and DCR (Additional file 1:FigureS5) CAF profiling and melanoma immunotherapy outcome, corroboratedthese findings andrevealeda similar however, there are a number of limitations. The most lack of association with RECIST. significant limitation is the fact that predictive bio- markers strictly require statistical proof by a test for Survival outcome and CAF parameters interaction in a randomized placebo-controlled trial, For survival analysis, the continuous CAF parameters which is no longer ethically possible for melanoma after were dichotomized into low and high statuses using the the approval of immune checkpoint therapy. Conse- Lausen and Schumacher method of maximally selected quently, all post-trial predictive biomarker studies are rank statistics for the standardized derivation of objective limited by the same statistical requirement. Instead, we thresholds from the population data (Additional file 1: analyzed an anti-PD-1 treated melanoma cohort and a Figure S6) [23]. In Cox regressions, both high Thy1 cell historical cohort predating immunotherapy to show a count and high FAP cell count were associated with pro- specific association between the biomarker and treat- longed PFS, whereas low SMA cell count was associated ment outcome. Indicative value is inferred if the bio- with prolonged PFS (Fig. 3aand Table 2). Similarly, OS marker is associated with outcome in the treated cohort had significant positive associations with both Thy1 and but a similar association is not observed in the control FAP cell counts, and a negative association with SMA cell cohort. This is best demonstrated in Fig. 3 and Table 3, count, which were specific to anti-PD-1 treated melanoma where the OS association with FAP undergoes a striking patients (all P <0.003; Fig. 3aand Table 3). To determine inversion as a function of presence or absence of anti- this distinction, a control melanoma cohort predating im- PD-1 therapy. Therefore, FAP has indicative value and munotherapy with known survival outcome was assessed may have future potential in a clinical assay to determine Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 5 of 10 SMA+FAP LO SMA HI FAP HI DAPI S100+HMB45 Thy1 SMA FAP Fig. 1 Cancer-associated fibroblast profiling by multiplex immunofluorescence in melanoma. Representative multispectral immunofluorescence images of CAF (Thy1, SMA, FAP) profiling in melanoma (magnification × 200; scale bar = 100 μm) (a), and corresponding visualizations of each CAF marker with nuclei (DAPI) and melanoma cells (S100 and HMB45) for the regions indicated (b). Abbreviations: CAF, cancer-associated fibroblast; DAPI, 4′,6-diamidino-2-phenylindole; HI, high; LO, low likelihood of survival benefit from anti-PD-1 therapy for patients [6]. Although our CAF profiling methodologies melanoma. Another limitation is the fact that this is a used quantitative fluorescence imaging systems for in- single-institutional retrospective study with a modest creased accuracy, the concept and design may be adapted sample size, even though all available relevant cases at for implementation on conventional pathology platforms Yale were collected at the time of the study. We look (for example, see Hartman et al. [28]). forward to prospective investigation of these assays or Recent studies indicate that mesenchymal or stromal similar in future clinical trials, especially since PD-1 abundance influences immunotherapy outcome [29, 30]. blockade is now widely used in the adjuvant setting However, the stromal compartment is heterogeneous where benefit is seen in only 1 in 5 treated melanoma and different CAF subsets may have divergent effects. In FAP SMA Thy1 Composite Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 6 of 10 50 50 50 R² = 0.0166 R² = 0.0425 R² = 0.0599 40 40 40 30 30 30 20 20 20 10 10 10 0 0 0 0 20406080 0 2040 6080 100 020 40 60 80 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) 40 40 40 R² = 0.0287 R² = 0.0439 R² = 0.0021 30 30 30 20 20 20 10 10 10 0 0 0 0 20406080 0 2040 6080 100 020 40 60 80 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) 30 30 30 R² = 0.0258 R² = 0.0332 R² = 0.0031 25 25 25 20 20 20 15 15 15 10 10 10 5 5 5 0 0 0 0 20406080 0 2040 6080 100 020 40 60 80 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) 25 25 25 R² = 2E-07 R² = 0.0003 R² = 0.0737 20 20 20 15 15 15 10 10 10 5 5 5 0 0 0 0 20406080 0 2040 6080 100 020 40 60 80 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) 80000 80000 80000 R² = 0.0012 R² = 0.0644 R² = 0.0152 60000 60000 60000 40000 40000 40000 20000 20000 20000 0 0 0 020 40 60 80 0 2040 6080 100 020 40 60 80 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) R² = 0.0034 R² = 0.0934 4000 R² = 0.0055 4000 4000 3000 3000 3000 2000 2000 2000 1000 1000 1000 0 0 0 0 2040 6080 0 20 406080 100 0 20 406080 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) Fig. 2 Immune markers and CAF parameters by cell counts in melanoma. Relationships between CAF (Thy1, SMA, FAP) markers and CD3 (a), CD4 (b), CD8 (c), CD20 (d), CD68 (e) and PD-L1 (f) in melanoma. Abbreviations: AU, arbitrary units; CAF, cancer-associated fibroblast; QIF, quantitative immunofluorescence the present study, the CAF population was stratified in activities, which may inform the design of new diagnos- terms of their expression of Thy1, SMA, and FAP. The tic and therapeutic strategies. differences in survival associations for SMA and FAP The intriguing role of FAP as a negative prognostic may reflect the functional complexity of CAF subsets. and positive indicative biomarker in melanoma is dem- According to a single-cell RNA sequencing study, up to onstrated by its positive association with survival out- seven CAF subsets with unique expression phenotypes come of anti-PD-1 treated melanoma patients, and its may exist in non-small cell lung cancer [9]. The identifi- inverse association with prognosis in the absence of im- cation of specific CAF subpopulations provides a foun- munotherapy. This is reminiscent of the well-known be- dation for future studies to deconvolute their specialized havior of HER2 as a negative prognostic and positive + + + + PD-L1 QIF (AU) CD68 QIF (AU) CD20 /total cells (%) CD8 /total cells (%) CD4 /total cells (%) CD3 /total cells (%) DAPI DAPI + + + + PD-L1 QIF (AU) CD68 QIF (AU) CD20 /total cells (%) CD8 /total cells (%) CD4 /total cells (%) CD3 /total cells (%) DAPI DAPI + + + + PD-L1 QIF (AU) CD68 QIF (AU) CD20 /total cells (%) CD8 /total cells (%) CD4 /total cells (%) CD3 /total cells (%) DAPI DAPI Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 7 of 10 + + + + + Thy1 /total SMA /Thy1 FAP /Thy1 100 100 100 50 50 50 0 0 0 0 500 1000 1500 2000 0 500 1000 1500 2000 0 500 1000 1500 2000 Time (days) Time (days) Time (days) No. at risk (P = 0.0072) No. at risk (P = 0.026) No. at risk (P = 0.017) HI 101 34 15 3 HI 24 4 0 0 HI 54 22 12 3 LO 16 3 1 1 LO 93 33 16 4 LO 63 15 4 1 No. at risk (P < 0.0001) No. at risk (P = 0.0008) No. at risk (P < 0.0001) HI 101 65 26 8 HI 24 10 1 0 HI 54 36 18 4 LO 16 5 1 1 LO 93 60 26 9 LO 63 34 9 5 No. at risk (P = 0.097) No. at risk (P = 0.038) No. at risk (P = 0.0090) HI 175 42 10 HI 10 3 0 HI 60 13 2 LO 19 7 0 LO 184 46 10 LO 134 36 8 Fig. 3 CAF parameters by cell counts and survival of melanoma patients treated with anti-PD-1 therapy and control melanoma patients. Kaplan–Meier analysis of progression-free survival and overall survival of anti-PD-1 treated melanoma patients (a) and overall survival of control melanoma patients (b) according to CAF (Thy1, SMA, FAP) parameters by cell counts. Low and high statuses were objectively defined using thresholds determined by maximally selected rank statistics (see Methods). Abbreviations: CAF, cancer-associated fibroblast; HI, high; LO, low predictive biomarker in breast cancer. Whereas HER2 is tumors [31]. This CAF-mediated mechanism may ex- the therapeutic target in the case of breast cancer, the plain the observed associations with survival benefit in role of FAP in immunotherapy is not well understood. anti-PD-1 therapy, and poor prognosis in the absence of The specific association of FAP with anti-PD-1 survival immunotherapy. Furthermore, our data demonstrate that advantage suggests mechanistic involvement. Recent these biomarkers are associated with survival outcome but supporting evidence has been published showing direct not RECIST-based response, which are different clinical interactions between CAFs and T cells, mediated endpoints. Multivariable analyses provided unique insights through coincident upregulation and engagement of PD- including the non-redundant role of FAP in the observed 1 on T cells, to drive T cell dysfunction and death within outcome associations when Thy1 and/or SMA are also Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 8 of 10 Table 2 Univariable and multivariable Cox regression analyses for progression-free survival of melanoma patients and CAF parameters by cell counts Variable Anti-PD-1 PFS (LO/HI) a a Univariable analysis Multivariable analysis per variable Multivariable analysis with Thy1, SMA, FAP HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value Thy1 /total 2.18 (1.17–3.81) 0.016 2.34 (1.21–4.28) 0.013 1.90 (0.98–3.48) 0.058 + + SMA /Thy1 0.55 (0.32–0.97) 0.038 0.55 (0.32–0.99) 0.048 0.71 (0.40–1.31) 0.26 + + FAP /Thy1 1.77 (1.11–2.89) 0.017 2.08 (1.28–3.44) 0.0030 1.79 (1.06–3.04) 0.031 Abbreviations: CAF cancer-associated fibroblast, CI confidence interval, HI high, HR hazard ratio, LO low, PFS progression-free survival Cox proportional hazards model included age, sex, mutation status, stage, treatment, and prior immune checkpoint blockade as covariates included in the Cox models. Melanoma mutation status melanoma immunotherapy outcome. The finding that was not associated with any CAF parameter [32]. The FAP is a negative prognostic but positive indicative CAF parameters also did not correlate with immune biomarker suggests mechanistic involvement and war- markers, which indicates independence of those meas- rants further study. Multiplex CAF profiling has the urement variables and non-redundancy, and may there- potential for application as a companion diagnostic in fore be complementary to existing biomarkers such as digital precision immuno-oncology and may be com- CD8 and PD-L1 [33, 34]. A combination biomarker strat- plementary to existing immune-related markers for egy is being studied to determine if combinations of CAF patient stratification. parameters with immune cell parameters have stronger associations with immunotherapy outcome. A predictive signature classifier computed from all available tissue data Conclusions is also under consideration. This study examines the clinical significance of The use of two independent image analysis tech- cancer-associated fibroblast (Thy1, SMA, FAP) pro- nologies to assess biomarkers and the concordant files in pretreatment tumor specimens to determine results from cell counts and QIF adds confidence in their association with immunotherapy outcome in thefindings. TheAQUA methodofQIF measures melanoma. We find that FAP, by both digital cell protein expression as cumulative signal intensity per counts and quantitative immunofluorescence of pro- unit compartment area, and it has been shown to be tein expression, shows significant positive associa- proportional to analyte concentration [35]. This is tions with survival outcome. The positive association fundamentally different from counts of digitally phe- is independent of baseline variables in multivariable notyped cells [36]. The similar results of the two analyses. In contrast, FAP is inversely associated with methodologies suggest shared biological relevance. prognosis in the absence of immunotherapy in a his- However, cell counts use intuitive absolute units torical cohort. The novel discovery that FAP is a and exhibited stronger associations with survival negative prognostic and positive indicative biomarker outcome than QIF, therefore, it may have a greater in melanoma suggests mechanistic involvement in potential for clinical translation in digital precision anti-PD-1 survival advantage. Its independence from immuno-oncology. previously described biomarkers like CD8 and PD-L1 In summary, this study demonstrates that pretreat- suggestitcould have valueincombinationwith ment CAF profiles, by in situ cell counts or QIF those markers to more accurately predict outcome to protein expression, are independently associated with immunotherapy. Table 3 Univariable and multivariable Cox regression analyses for overall survival of melanoma patients and CAF parameters by cell counts Variable Control OS Anti-PD-1 OS (LO/HI) a a Univariable analysis Univariable analysis Multivariable analysis per variable Multivariable analysis with Thy1, SMA, FAP HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value Thy1 /total 1.65 (0.87–2.88) 0.12 4.66 (2.34–8.82) < 0.0001 4.67 (2.19–9.53) 0.0001 3.02 (1.44–6.10) 0.0044 + + SMA /Thy1 0.45 (0.22–1.07) 0.070 0.34 (0.18–0.68) 0.0029 0.32 (0.16–0.67) 0.0027 0.62 (0.30–1.31) 0.20 + + FAP /Thy1 0.57 (0.37–0.88) 0.012 4.11 (2.05–9.14) < 0.0001 4.64 (2.27–10.52) < 0.0001 3.61 (1.65–8.56) 0.0011 Abbreviations: CAF cancer-associated fibroblast, CI confidence interval, HI high, HR hazard ratio, LO low, OS overall survival Cox proportional hazards model included age, sex, mutation status, stage, treatment, and prior immune checkpoint blockade as covariates Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 9 of 10 Additional file Received: 21 January 2019 Accepted: 11 July 2019 Additional file 1: Supplementary Figure 1. Linear regressions of CAF parameters in melanoma by cell counts and quantitative References immunofluorescence. Correlation between cell counts and QIF scores for 1. Topalian SL, Hodi FS, Brahmer JR, et al. Safety, activity, and immune CAF (Thy1, SMA, FAP) markers (A). Relationships between Thy1, SMA, and correlates of anti-PD-1 antibody in cancer. N Engl J Med. 2012;366(26): FAP by cell counts and QIF (B). Abbreviations: AU, arbitrary units; CAF, 2443–54. cancerassociated fibroblast; QIF, quantitative immunofluorescence. (PDF 2. Weber JS, D'Angelo SP, Minor D, et al. Nivolumab versus chemotherapy in 439 kb) patients with advanced melanoma who progressed after anti-CTLA-4 treatment (CheckMate 037): a randomised, controlled, open-label, phase 3 trial. Lancet Oncol. 2015;16(4):375–84. Acknowledgements 3. Wolchok JD, Chiarion-Sileni V, Gonzalez R, et al. Overall survival with The authors thank Lori A. Charette and the staff of Yale Pathology Tissue combined Nivolumab and Ipilimumab in advanced melanoma. N Engl J Services for expert histology services. This work is based on the PhD Med. 2017;377(14):1345–56. dissertation research of Dr. Pok Fai Wong as a Gruber Science Fellow at 4. Robert C, Schachter J, Long GV, et al. Pembrolizumab versus Ipilimumab in Yale University. advanced melanoma. N Engl J Med. 2015;372(26):2521–32. 5. Ribas A, Hamid O, Daud A, et al. Association of Pembrolizumab with Tumor Response and Survival among Patients with Advanced Melanoma. JAMA. Authors’ contributions 2016;315(15):1600–9. Conception and design: PFW, DLR. Case selection and clinical data collection: 6. Eggermont AMM, Blank CU, Mandala M, et al. Adjuvant Pembrolizumab PFW, JWS, HMK, DLR. Multiplex cancer-associated fibroblast analysis: PFW. versus placebo in resected stage III melanoma. N Engl J Med. 2018;378(19): Multiplex immune marker analysis: PFW, SG. Statistical analysis: PFW, WW, DZ. 1789–801. Drafting of manuscript: PFW, DLR. Critical revision of manuscript: All authors. 7. Abdel-Wahab N, Shah M, Suarez-Almazor ME. Adverse events associated Final approval of manuscript: All authors. Study supervision: DLR. with immune checkpoint blockade in patients with Cancer: a systematic review of case reports. PLoS One. 2016;11(7):e0160221. 8. Tirosh I, Izar B, Prakadan SM, et al. Dissecting the multicellular ecosystem of Funding metastatic melanoma by single-cell RNA-seq. Science. 2016;352(6282):189–96. This work was supported by funds from Navigate BioPharma (Novartis subsidiary), 9. Lambrechts D, Wauters E, Boeckx B, et al. Phenotype molding of stromal Yale SPORE in Lung Cancer and Yale Cancer Center to D.L. Rimm; R01 CA227473, cells in the lung tumor microenvironment. Nat Med. 2018;24(8):1277–89. K24CA172123 and P50 CA121974 to H.M. Kluger; and the Gruber Science 10. Fiore VF, Strane PW, Bryksin AV, White ES, Hagood JS, Barker TH. Conformational Fellowship to P.F. Wong from the Gruber Foundation. The funding coupling of integrin and Thy-1 regulates Fyn priming and fibroblast sources had no role in study design; collection, analysis and interpretation of mechanotransduction. J Cell Biol. 2015;211(1):173–90. data; preparation of the manuscript or the decision to submit for publication. 11. Hinz B, Celetta G, Tomasek JJ, Gabbiani G, Chaponnier C. Alpha-smooth muscle actin expression upregulates fibroblast contractile activity. Mol Biol Cell. 2001;12(9):2730–41. Availability of data and materials 12. Wong PF, Gall MG, Bachovchin WW, McCaughan GW, Keane FM, Gorrell MD. De-identified datasets used and/or analyzed during the current study are Neuropeptide Y is a physiological substrate of fibroblast activation protein: available from the corresponding author upon reasonable request. enzyme kinetics in blood plasma and expression of Y2R and Y5R in human liver cirrhosis and hepatocellular carcinoma. Peptides. 2016;75:80–95. 13. Garin-Chesa P, Old LJ, Rettig WJ. Cell surface glycoprotein of reactive Ethics approval and consent to participate stromal fibroblasts as a potential antibody target in human epithelial The study was approved by the Yale Human Investigation Committee protocol cancers. Proc Natl Acad Sci U S A. 1990;87(18):7235–9. #9505008219 and conducted in accordance with the Declaration of Helsinki. All 14. Feig C, Jones JO, Kraman M, et al. Targeting CXCL12 from FAP-expressing patients provided written informed consent or waiver of consent in carcinoma-associated fibroblasts synergizes with anti-PD-L1 immunotherapy circumstances stipulated by the HIC protocol. in pancreatic cancer. Proc Natl Acad Sci U S A. 2013;110(50):20212–7. 15. Ozdemir BC, Pentcheva-Hoang T, Carstens JL, et al. Depletion of carcinoma- associated fibroblasts and fibrosis induces immunosuppression and Consent for publication accelerates pancreas cancer with reduced survival. Cancer Cell. 2014; Not applicable. 25(6):719–34. 16. Kumar V, Donthireddy L, Marvel D, et al. Cancer-associated fibroblasts neutralize the anti-tumor effect of CSF1 receptor blockade by inducing PMN-MDSC infiltration of tumors. Cancer Cell. 2017;32(5):654–668.e655. Competing interests 17. Wong PF, Wei W, Smithy JW, et al. Multiplex quantitative analysis of tumor- D.L. Rimm has served as a consultant, advisor or served on a Scientific infiltrating lymphocytes and immunotherapy outcome in metastatic Advisory Board for Amgen, Astra Zeneca, Agendia, Biocept, Bristol-Meyers melanoma. Clin Cancer Res. 2019;25(8):2442-49. Squibb, Cell Signaling Technology, Cepheid, Daiichi Sankyo, GSK, Merck, 18. Qin Y, Petaccia de Macedo M, Reuben A, et al. Parallel profiling of immune NanoString, PerkinElmer, PAIGE, and Ultivue. He has received research fund- infiltrate subsets in uveal melanoma versus cutaneous melanoma unveils ing from Astra Zeneca, Cepheid, Navigate/Novartis, NextCure, Lilly, Ultivue, similarities and differences: a pilot study. Oncoimmunology. 2017;6(6): and PerkinElmer. H.M. Kluger has served as a consultant for Corvus, Nektar, e1321187. Biodesix, Genentech, Pfizer, Merck and Celldex, and has received research support from Merck, Apexigen and Bristol-Meyers Squibb. All other authors 19. Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria have no competing interests. in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer. 2009; 45(2):228–47. Author details 20. Brown JR, Wimberly H, Lannin DR, Nixon C, Rimm DL, Bossuyt V. Department of Pathology, Yale School of Medicine, New Haven, CT 06510, Multiplexed quantitative analysis of CD3, CD8, and CD20 predicts USA. Yale Cancer Center, Yale School of Medicine, New Haven, CT 06510, response to neoadjuvant chemotherapy in breast cancer. Clin Cancer USA. Department of Biostatistics, Yale School of Public Health, New Haven, Res. 2014;20(23):5995–6005. CT 06510, USA. Section of Medical Oncology, Department of Internal 21. Huang W, Hennrick K,DrewS.Acolorful future of quantitative Medicine, Yale School of Medicine, New Haven, CT 06510, USA. Department pathology: validation of Vectra technology using chromogenic of Pathology, Yale School of Medicine, 310 Cedar St, BML 116, PO Box multiplexed immunohistochemistry and prostate tissue microarrays. 208023, New Haven, CT 06520, USA. Hum Pathol. 2013;44(1):29–38. Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 10 of 10 22. Camp RL, Chung GG, Rimm DL. Automated subcellular localization and quantification of protein expression in tissue microarrays. Nat Med. 2002; 8(11):1323–7. 23. Lausen B, Schumacher M. Maximally selected rank statistics. Biometrics. 1992;48(1):73–85. 24. Eton O, Legha SS, Moon TE, et al. Prognostic factors for survival of patients treated systemically for disseminated melanoma. J Clin Oncol. 1998;16(3): 1103–11. 25. Manola J, Atkins M, Ibrahim J, Kirkwood J. Prognostic factors in metastatic melanoma: a pooled analysis of eastern cooperative oncology group trials. J Clin Oncol. 2000;18(22):3782–93. 26. Neuman HB, Patel A, Ishill N, et al. A single-institution validation of the AJCC staging system for stage IV melanoma. Ann Surg Oncol. 2008;15(7):2034–41. 27. Joosse A, Collette S, Suciu S, et al. Sex is an independent prognostic indicator for survival and relapse/progression-free survival in metastasized stage III to IV melanoma: a pooled analysis of five European organisation for research and treatment of cancer randomized controlled trials. J Clin Oncol. 2013;31(18): 2337–46. 28. Hartman DJ, Ahmad F, Ferris RL, Rimm DL, Pantanowitz L. Utility of CD8 score by automated quantitative image analysis in head and neck squamous cell carcinoma. Oral Oncol. 2018;86:278–87. 29. Hugo W, Zaretsky JM, Sun L, et al. Genomic and transcriptomic features of response to anti-PD-1 therapy in metastatic melanoma. Cell. 2016;165(1):35–44. 30. Zhao F, Evans K, Xiao C, et al. Stromal fibroblasts mediate anti-PD-1 resistance via MMP-9 and dictate TGFbeta inhibitor sequencing in melanoma. Cancer Immunol Res. 2018;6(12):1459–71. 31. Lakins MA, Ghorani E, Munir H, Martins CP, Shields JD. Cancer-associated fibroblasts induce antigen-specific deletion of CD8+ T cells to protect tumour cells. Nat Commun. 2018;9(1):948. 32. Cancer Genome Atlas Network. Genomic classification of cutaneous melanoma. Cell. 2015;161(7):1681–96. 33. Tumeh PC, Harview CL, Yearley JH, et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature. 2014;515(7528):568–71. 34. Chen PL, Roh W, Reuben A, et al. Analysis of immune signatures in longitudinal tumor samples yields insight into biomarkers of response and mechanisms of resistance to immune checkpoint blockade. Cancer Discov. 2016;6(8):827–37. 35. Toki MI, Cecchi F, Hembrough T, Syrigos KN, Rimm DL. Proof of the quantitative potential of immunofluorescence by mass spectrometry. Lab Investig. 2017;97(3):329–34. 36. Carvajal-Hausdorf DE, Schalper KA, Neumeister VM, Rimm DL. Quantitative measurement of cancer tissue biomarkers in the lab and in the clinic. Lab Investig. 2015;95(4):385–96. Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal for ImmunoTherapy of Cancer Springer Journals

Multiplex quantitative analysis of cancer-associated fibroblasts and immunotherapy outcome in metastatic melanoma

Loading next page...
 
/lp/springer-journals/multiplex-quantitative-analysis-of-cancer-associated-fibroblasts-and-p0QFGPwbKK

References (41)

Publisher
Springer Journals
Copyright
Copyright © 2019 by The Author(s).
Subject
Medicine & Public Health; Oncology; Immunology
eISSN
2051-1426
DOI
10.1186/s40425-019-0675-0
Publisher site
See Article on Publisher Site

Abstract

Background: The cancer-associated fibroblast (CAF) population is implicated in immune dysregulation. Here, we test the hypothesis that CAF profiles in pretreatment tumor specimens are associated with response to immune checkpoint blockade of programmed cell death 1 (PD-1). Methods: Pretreatment whole tissue sections from 117 melanoma patients treated with anti-PD-1 therapy were assessed by multiplex immunofluorescence to detect CAFs defined by Thy1, smooth muscle actin (SMA), and fibroblast activation protein (FAP). Two independent image analysis technologies were used: inForm software (PerkinElmer) to quantify cell counts, and AQUA™ to measure protein by quantitative immunofluorescence (QIF). CAF parameters by both methodologies were assessed for association with previously measured immune markers (CD3, CD4, CD8, CD20, CD68, PD-L1), best overall response, progression-free survival (PFS), and overall survival (OS). Results: CAF parameters, by cell counts or QIF, did not correlate with immune markers nor with best overall response. However, both Thy1 and FAP cell counts had significant positive associations with PFS (all P < 0.05) and OS (all P < 0.003). SMA cell counts showed negative associations with outcome in anti-PD-1 treated patients. Similar associations were not observed in a control cohort of historical melanoma patients predating immunotherapy. Instead, FAP was a negative prognostic biomarker (P = 0.01) in the absence of immunotherapy. Multivariable analyses revealed significant PFS and OS associations with the CAF parameters were independent of baseline variables. Conclusions: Pretreatment CAF profiles are associated with melanoma immunotherapy outcome. Multiplex CAF analysis has potential as an objective companion diagnostic in immuno-oncology. Keywords: Biomarkers, Cancer-associated fibroblasts, Fibroblast activation protein, Immunotherapy, Melanoma, PD-1 Introduction protein 4 (CTLA-4), which is targeted by ipilimumab [4]. Immune checkpoint blockade has become a new stand- Nevertheless, clinical benefit is limited to ~ 40% of meta- ard in melanoma immunotherapy and the overall sur- static melanoma patients treated with anti-PD-1 therapy, vival of patients with metastatic disease has improved which is compounded by the lack of approved predictive from ~ 9 months before 2011 to greater than 3 years strategies [1, 5]. Due to widespread use of PD-1 blockade [1–3]. The tumor-infiltrating lymphocyte (TIL) popu- and its recent introduction into the adjuvant setting [6], lation expresses immune checkpoints, programmed there is an increasing need for robust biomarkers to in- cell death 1 (PD-1), which is targeted by pembrolizumab form the practice of precision immuno-oncology [7]. and nivolumab; and cytotoxic T-lymphocyte associated The cancer-associated fibroblast (CAF) population en- gages in a complex and poorly understood interplay with tumor cells and immune cells, and are the predominant * Correspondence: david.rimm@yale.edu stromal cell type within the tumor microenvironment. Department of Pathology, Yale School of Medicine, New Haven, CT 06510, CAFs are characterized by expression of Thy1, with sub- USA Yale Cancer Center, Yale School of Medicine, New Haven, CT 06510, USA sets expressing smooth muscle actin (SMA) or fibroblast Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 2 of 10 activation protein (FAP) [8, 9]. Thy1 is a glycophosphati- only included pretreatment formalin-fixed, paraffin- dylinositol (GPI)-anchored cell surface protein that binds embedded (FFPE) specimens after review by a board- to integrins and may be involved in cell–cell adhesion certified pathologist. All specimens were collected from [10]. SMA is a major component of the contractile ap- the Yale Pathology archives. Clinicopathological data paratus that allows fibroblasts to produce contractile were collected from clinical records and pathology re- force [11]. FAP is a type II transmembrane serine prote- ports; the data cut-off date was September 1, 2017. Re- ase that cleaves collagen I as an endopeptidase and en- sponse Evaluation Criteria in Solid Tumors (RECIST) gages in post-translational modification of neuropeptide 1.1 were used to determine best overall response as Y as a dipeptidyl peptidase, which is the rare ability to complete response (CR), partial response (PR), stable hydrolyze the post-proline bond two residues from the disease (SD), or progressive disease (PD), and objective N-terminus of substrates [12]. FAP is weakly expressed response rate (ORR; CR/PR) and disease control rate or not detected in normal adult tissues but is upregu- (DCR; CR/PR/SD) [19]. A historical cohort of 194 mel- lated at sites of activated stroma in tumors and in anoma patients, collected prior to the advent of anti-PD- chronic inflammation [13]. Emerging preclinical evidence 1, was used as the control group. Cohort characteristics implicates CAFs in immune dysregulation and response are detailed in Table 1. Other characteristics of the anti- to immunotherapy [14–16]. However, CAFs represent a PD-1 treated cohort including the melanoma specimen, heterogeneous group and different CAF subsets may have time interval to anti-PD-1 therapy, and prior immune opposing functions. A more comprehensive understand- checkpoint blockade are shown in Additional file 1: ing of different CAF subsets as well as their impact on hu- Table S1. All patients provided written informed consent man immunotherapy outcome is needed. or waiver of consent. The study was approved by the Yale We hypothesized that pretreatment CAF profiles of Human Investigation Committee protocol #9505008219 patient tumors would be associated with immunotherapy and conducted in accordance with the Declaration of outcome. However, predictive biomarkers strictly require Helsinki. statistical evidence from a formal test for interaction in randomized placebo-controlled studies, which are no Multiplex immunofluorescence CAF panel longer ethically possible for melanoma. Therefore, we 5-plex immunofluorescence using isotype-specific anti- tested a control cohort of historical melanoma patients bodies was performed on FFPE whole tissue sections for predating immunotherapy instead to distinguish prog- simultaneous detection of markers as previously described nostic value and show a specific association between the [20]. The protocol is detailed in the Additional file 1. biomarker and treatment outcome. We describe this type of biomarker as “indicative”, a separate category Image analysis by two independent methods: cell counts from truly predictive biomarkers under existing statis- versus quantitative immunofluorescence tical definitions [17]. Briefly, indicative value is demon- Cell counts were determined by the pattern recognition strated when: [1] the hazard ratio is statistically software, inForm Tissue Finder (PerkinElmer, Waltham, significant in the treatment cohort and is not significant MA, USA), on multispectral images acquired using a in the control cohort; or [2] the hazard ratio is statisti- Vectra 3 system (PerkinElmer) as previously described cally significant in both the treatment and control co- [21]. Multispectral images were decomposed into their horts, but the respective 95% confidence intervals do not various components by spectral unmixing using a digital significantly overlap. The former characteristic is purely spectral library consisting of spectral profiles of each of indicative, and the latter is both prognostic and indica- the fluorophores. Automated tissue segmentation identi- tive [17]. fied tumor and stroma regions. Cell segmentation within Here, we assess the clinical significance of CAFs for these regions identified individual cells and respective the prediction of immunotherapy outcome in metastatic nuclei, cytoplasm, and membrane components using sig- melanoma. We hypothesize that the expression of these nal in the nucleus and membrane as internal and exter- candidate biomarkers, Thy1, SMA, and FAP, will classify nal cell borders, then cells were phenotyped for marker anti-PD-1 therapy treated patients into groups that expression. Cell counts for each melanoma case were benefit and those that do not. calculated in terms of the number of cells positive for the marker of interest as a percentage of the cell popula- Methods tion in which it was measured. Protein expression of the Patient cohort various markers was determined by the automated quan- The study cohort is a retrospective collection of 117 titative analysis (AQUA) method of QIF on fluorescence melanoma patients treated with anti-PD-1 therapy in the images acquired using a PM-2000 system (Navigate metastatic setting between 2011 and 2017 at Yale Cancer BioPharma, Carlsbad, CA, USA) as previously described Center. Uveal melanoma was excluded [18]. The analysis [22]. A total compartment, consisting of all cells, or a Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 3 of 10 Table 1 Clinicopathological characteristics of the melanoma cohort treated with anti-PD-1 therapy and the control melanoma cohort for CAF profiling Characteristic Anti-PD-1 patients, No. (%) Objective response Disease control rate Control patients, No. (%) rate (CR/PR), No. (%) (CR/PR/SD), No. (%) Overall 117 (100) 55 (47) 81 (69) 194 (100) Age (y) < 65 67 (57) 34 (51) 51 (76) 87 (45) ≥ 65 50 (43) 21 (42) 30 (60) 107 (55) Sex Male 70 (60) 35 (50) 48 (69) 110 (57) Female 47 (40) 20 (43) 33 (70) 84 (43) Treatment Pembrolizumab 41 (35) 20 (49) 30 (73) 0 Nivolumab 18 (15) 7 (39) 9 (50) 0 Ipilimumab plus nivolumab 58 (50) 28 (48) 42 (72) 0 Prior immune checkpoint blockade Yes 36 (31) 13 (36) 22 (61) 0 No 81 (69) 42 (52) 59 (73) 194 (100) Mutation status BRAF 39 (33) 19 (49) 27 (69) NA NRAS 18 (15) 8 (44) 11 (61) NA KIT 2 (2) 1 (50) 2 (100) NA None detected 58 (50) 27 (47) 41 (71) NA Stage at diagnosis I 24 (21) 14 (58) 19 (79) 77 (40) II 23 (20) 12 (52) 16 (70) 80 (41) III 38 (32) 16 (42) 24 (63) 30 (15) IV 20 (17) 6 (30) 13 (65) 3 (2) Not available 12 (10) 7 (58) 9 (75) 4 (2) Abbreviations: CAF cancer-associated fibroblast, CR complete response, NA not available, PR partial response, SD stable disease Thy1 compartment was generated by automated process- (PFS) and overall survival (OS) functions were computed ing and thresholding of the DAPI signal or Thy1 signal, and comparisons were made by the log-rank test. Multi- respectively. QIF scores were calculated by dividing the variable Cox proportional hazards models included age, summed pixel intensities for the marker of interest by the sex, mutation status, stage, treatment, and prior immune area of the compartment in which it was measured [22]. checkpoint blockade as covariates [24–27]. All statistical Overall QIF scores were derived for each melanoma case tests were two-sided and statistical significance was de- by averaging scores from each field of view. fined as P < 0.05. Statistical analysis was performed using GraphPad Prism 7 (GraphPad Software, La Jolla, CA, Statistical analysis USA) and JMP Pro 13 (SAS Institute, Cary, NC, USA). Statistical comparisons for cell count and QIF data were The sample size of 117 patients had at least 80% power made using unpaired t-test or analysis of variance at P = 0.05 to detect a difference in means of 0.52 stand- (ANOVA) followed by Tukey’s test for multiple compari- ard deviations in each CAF parameter for responders sons as appropriate. The Lausen and Schumacher (CR/PR) versus non-responders (SD/PD). method of maximally selected rank statistics, a powerful nonparametric method for assessing predictive power of Results a continuous variable for a dependent variable, was used Correlation between cell counts and quantitative to determine thresholds to objectively define low and immunofluorescence high statuses for the measured CAF parameters [23]. Tissue biomarkers can be quantified in situ by counting Kaplan–Meier estimates of progression-free survival positive cells for the biomarker or in terms of quantitative Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 4 of 10 protein expression levels per unit area. These are two in- for prognostic value in place of a placebo arm. Simi- dependent types of parameters and may be nonequivalent lar associations were not observed in the control pa- in clinical significance. The relationship between cell tients (Fig. 3band Table 3). Remarkably, FAP cell count counts and QIF was assessed by linear regression, which was a significant negative prognostic biomarker in the ab- revealed a positive correlation for Thy1 (R = 0.35), SMA sence of immunotherapy (P = 0.01) with an inverted haz- 2 2 (R = 0.36), and FAP (R = 0.62) (Additional file 1: ard ratio (HR = 0.57, 95% CI, 0.37–0.88) relative to that of Figure S1A). On the contrary, there was no correlation be- the anti-PD-1 patients (HR = 4.11, 95% CI, 2.05–9.14) tween different markers, which confirmed their independ- (Table 3). Multivariable analyses further revealed signifi- ence (Additional file 1:Figure S1B). cant survival associations with the CAF parameters, par- ticularly for FAP, independent of age, sex, mutation, stage, Immune markers and CAF parameters treatment, and prior immune checkpoint blockade Pretreatment whole tissue sections from 117 melanoma (Tables 2–3). The QIF data showed similar trends in patients treated with anti-PD-1 therapy underwent CAF relation to survival (Additional file 1:FigureS7and (Thy1, SMA, FAP) profiling by multiplex immunofluor- Additional file 1:TablesS2–S3). Survival analysis by escence (Fig. 1). The relationship between CAFs and in- treatment group generally showed similar trends despite filtration of immune cell populations or expression of the reduction in statistical power (Additional file 1: immune markers in melanoma was assessed by linear Tables S4–S5). regression with previous data [17]. There was no cor- relation between the CAF parameters and CD3, CD4, Discussion CD8, CD20, CD68, or PD-L1, which confirmed their In this study, we determine the clinical significance of independence of those immune markers (Fig. 2 and pretreatment CAF (Thy1, SMA, FAP) profiles according Additional file 1: Figure S2). to both in situ cell counts and QIF protein expression in relation to immunotherapy outcome in metastatic mel- Best overall response by RECIST and CAF parameters anoma. PFS and OS had positive associations with Thy1 The CAF parameters, by cell counts or QIF, were ana- and FAP cell counts, and negative associations with lyzed in relation to specimen-specific variables and SMA cell count, which were specific to anti-PD-1 tumor burden classifications defined by RECIST 1.1 treated patients. Significant PFS and OS associations [19]. There were no significant associations with sex or with the CAF parameters were independent of age, sex, mutation status of melanoma patients for the CAF pa- mutation, stage, treatment, and prior immune check- rameters by cell counts or QIF (all P > 0.05; Additional point blockade [24–27]. While the two quantitative file 1: Figure S3). Neither Thy1, SMA, nor FAP cell methods are independent, cell counts correlated with counts were associated with best overall response defined by QIF and revealed concordant associations with response RECIST (all P > 0.05; Additional file 1:FigureS4A). The cor- and survival outcome. responding QIF data (Additional file 1:FigureS4B)and fur- This study attempts to rigorously investigate multiplex ther analyses on ORR and DCR (Additional file 1:FigureS5) CAF profiling and melanoma immunotherapy outcome, corroboratedthese findings andrevealeda similar however, there are a number of limitations. The most lack of association with RECIST. significant limitation is the fact that predictive bio- markers strictly require statistical proof by a test for Survival outcome and CAF parameters interaction in a randomized placebo-controlled trial, For survival analysis, the continuous CAF parameters which is no longer ethically possible for melanoma after were dichotomized into low and high statuses using the the approval of immune checkpoint therapy. Conse- Lausen and Schumacher method of maximally selected quently, all post-trial predictive biomarker studies are rank statistics for the standardized derivation of objective limited by the same statistical requirement. Instead, we thresholds from the population data (Additional file 1: analyzed an anti-PD-1 treated melanoma cohort and a Figure S6) [23]. In Cox regressions, both high Thy1 cell historical cohort predating immunotherapy to show a count and high FAP cell count were associated with pro- specific association between the biomarker and treat- longed PFS, whereas low SMA cell count was associated ment outcome. Indicative value is inferred if the bio- with prolonged PFS (Fig. 3aand Table 2). Similarly, OS marker is associated with outcome in the treated cohort had significant positive associations with both Thy1 and but a similar association is not observed in the control FAP cell counts, and a negative association with SMA cell cohort. This is best demonstrated in Fig. 3 and Table 3, count, which were specific to anti-PD-1 treated melanoma where the OS association with FAP undergoes a striking patients (all P <0.003; Fig. 3aand Table 3). To determine inversion as a function of presence or absence of anti- this distinction, a control melanoma cohort predating im- PD-1 therapy. Therefore, FAP has indicative value and munotherapy with known survival outcome was assessed may have future potential in a clinical assay to determine Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 5 of 10 SMA+FAP LO SMA HI FAP HI DAPI S100+HMB45 Thy1 SMA FAP Fig. 1 Cancer-associated fibroblast profiling by multiplex immunofluorescence in melanoma. Representative multispectral immunofluorescence images of CAF (Thy1, SMA, FAP) profiling in melanoma (magnification × 200; scale bar = 100 μm) (a), and corresponding visualizations of each CAF marker with nuclei (DAPI) and melanoma cells (S100 and HMB45) for the regions indicated (b). Abbreviations: CAF, cancer-associated fibroblast; DAPI, 4′,6-diamidino-2-phenylindole; HI, high; LO, low likelihood of survival benefit from anti-PD-1 therapy for patients [6]. Although our CAF profiling methodologies melanoma. Another limitation is the fact that this is a used quantitative fluorescence imaging systems for in- single-institutional retrospective study with a modest creased accuracy, the concept and design may be adapted sample size, even though all available relevant cases at for implementation on conventional pathology platforms Yale were collected at the time of the study. We look (for example, see Hartman et al. [28]). forward to prospective investigation of these assays or Recent studies indicate that mesenchymal or stromal similar in future clinical trials, especially since PD-1 abundance influences immunotherapy outcome [29, 30]. blockade is now widely used in the adjuvant setting However, the stromal compartment is heterogeneous where benefit is seen in only 1 in 5 treated melanoma and different CAF subsets may have divergent effects. In FAP SMA Thy1 Composite Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 6 of 10 50 50 50 R² = 0.0166 R² = 0.0425 R² = 0.0599 40 40 40 30 30 30 20 20 20 10 10 10 0 0 0 0 20406080 0 2040 6080 100 020 40 60 80 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) 40 40 40 R² = 0.0287 R² = 0.0439 R² = 0.0021 30 30 30 20 20 20 10 10 10 0 0 0 0 20406080 0 2040 6080 100 020 40 60 80 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) 30 30 30 R² = 0.0258 R² = 0.0332 R² = 0.0031 25 25 25 20 20 20 15 15 15 10 10 10 5 5 5 0 0 0 0 20406080 0 2040 6080 100 020 40 60 80 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) 25 25 25 R² = 2E-07 R² = 0.0003 R² = 0.0737 20 20 20 15 15 15 10 10 10 5 5 5 0 0 0 0 20406080 0 2040 6080 100 020 40 60 80 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) 80000 80000 80000 R² = 0.0012 R² = 0.0644 R² = 0.0152 60000 60000 60000 40000 40000 40000 20000 20000 20000 0 0 0 020 40 60 80 0 2040 6080 100 020 40 60 80 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) R² = 0.0034 R² = 0.0934 4000 R² = 0.0055 4000 4000 3000 3000 3000 2000 2000 2000 1000 1000 1000 0 0 0 0 2040 6080 0 20 406080 100 0 20 406080 100 + + + + + Thy1 /total cells (%) SMA /Thy1 cells (%) FAP /Thy1 cells (%) Fig. 2 Immune markers and CAF parameters by cell counts in melanoma. Relationships between CAF (Thy1, SMA, FAP) markers and CD3 (a), CD4 (b), CD8 (c), CD20 (d), CD68 (e) and PD-L1 (f) in melanoma. Abbreviations: AU, arbitrary units; CAF, cancer-associated fibroblast; QIF, quantitative immunofluorescence the present study, the CAF population was stratified in activities, which may inform the design of new diagnos- terms of their expression of Thy1, SMA, and FAP. The tic and therapeutic strategies. differences in survival associations for SMA and FAP The intriguing role of FAP as a negative prognostic may reflect the functional complexity of CAF subsets. and positive indicative biomarker in melanoma is dem- According to a single-cell RNA sequencing study, up to onstrated by its positive association with survival out- seven CAF subsets with unique expression phenotypes come of anti-PD-1 treated melanoma patients, and its may exist in non-small cell lung cancer [9]. The identifi- inverse association with prognosis in the absence of im- cation of specific CAF subpopulations provides a foun- munotherapy. This is reminiscent of the well-known be- dation for future studies to deconvolute their specialized havior of HER2 as a negative prognostic and positive + + + + PD-L1 QIF (AU) CD68 QIF (AU) CD20 /total cells (%) CD8 /total cells (%) CD4 /total cells (%) CD3 /total cells (%) DAPI DAPI + + + + PD-L1 QIF (AU) CD68 QIF (AU) CD20 /total cells (%) CD8 /total cells (%) CD4 /total cells (%) CD3 /total cells (%) DAPI DAPI + + + + PD-L1 QIF (AU) CD68 QIF (AU) CD20 /total cells (%) CD8 /total cells (%) CD4 /total cells (%) CD3 /total cells (%) DAPI DAPI Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 7 of 10 + + + + + Thy1 /total SMA /Thy1 FAP /Thy1 100 100 100 50 50 50 0 0 0 0 500 1000 1500 2000 0 500 1000 1500 2000 0 500 1000 1500 2000 Time (days) Time (days) Time (days) No. at risk (P = 0.0072) No. at risk (P = 0.026) No. at risk (P = 0.017) HI 101 34 15 3 HI 24 4 0 0 HI 54 22 12 3 LO 16 3 1 1 LO 93 33 16 4 LO 63 15 4 1 No. at risk (P < 0.0001) No. at risk (P = 0.0008) No. at risk (P < 0.0001) HI 101 65 26 8 HI 24 10 1 0 HI 54 36 18 4 LO 16 5 1 1 LO 93 60 26 9 LO 63 34 9 5 No. at risk (P = 0.097) No. at risk (P = 0.038) No. at risk (P = 0.0090) HI 175 42 10 HI 10 3 0 HI 60 13 2 LO 19 7 0 LO 184 46 10 LO 134 36 8 Fig. 3 CAF parameters by cell counts and survival of melanoma patients treated with anti-PD-1 therapy and control melanoma patients. Kaplan–Meier analysis of progression-free survival and overall survival of anti-PD-1 treated melanoma patients (a) and overall survival of control melanoma patients (b) according to CAF (Thy1, SMA, FAP) parameters by cell counts. Low and high statuses were objectively defined using thresholds determined by maximally selected rank statistics (see Methods). Abbreviations: CAF, cancer-associated fibroblast; HI, high; LO, low predictive biomarker in breast cancer. Whereas HER2 is tumors [31]. This CAF-mediated mechanism may ex- the therapeutic target in the case of breast cancer, the plain the observed associations with survival benefit in role of FAP in immunotherapy is not well understood. anti-PD-1 therapy, and poor prognosis in the absence of The specific association of FAP with anti-PD-1 survival immunotherapy. Furthermore, our data demonstrate that advantage suggests mechanistic involvement. Recent these biomarkers are associated with survival outcome but supporting evidence has been published showing direct not RECIST-based response, which are different clinical interactions between CAFs and T cells, mediated endpoints. Multivariable analyses provided unique insights through coincident upregulation and engagement of PD- including the non-redundant role of FAP in the observed 1 on T cells, to drive T cell dysfunction and death within outcome associations when Thy1 and/or SMA are also Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 8 of 10 Table 2 Univariable and multivariable Cox regression analyses for progression-free survival of melanoma patients and CAF parameters by cell counts Variable Anti-PD-1 PFS (LO/HI) a a Univariable analysis Multivariable analysis per variable Multivariable analysis with Thy1, SMA, FAP HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value Thy1 /total 2.18 (1.17–3.81) 0.016 2.34 (1.21–4.28) 0.013 1.90 (0.98–3.48) 0.058 + + SMA /Thy1 0.55 (0.32–0.97) 0.038 0.55 (0.32–0.99) 0.048 0.71 (0.40–1.31) 0.26 + + FAP /Thy1 1.77 (1.11–2.89) 0.017 2.08 (1.28–3.44) 0.0030 1.79 (1.06–3.04) 0.031 Abbreviations: CAF cancer-associated fibroblast, CI confidence interval, HI high, HR hazard ratio, LO low, PFS progression-free survival Cox proportional hazards model included age, sex, mutation status, stage, treatment, and prior immune checkpoint blockade as covariates included in the Cox models. Melanoma mutation status melanoma immunotherapy outcome. The finding that was not associated with any CAF parameter [32]. The FAP is a negative prognostic but positive indicative CAF parameters also did not correlate with immune biomarker suggests mechanistic involvement and war- markers, which indicates independence of those meas- rants further study. Multiplex CAF profiling has the urement variables and non-redundancy, and may there- potential for application as a companion diagnostic in fore be complementary to existing biomarkers such as digital precision immuno-oncology and may be com- CD8 and PD-L1 [33, 34]. A combination biomarker strat- plementary to existing immune-related markers for egy is being studied to determine if combinations of CAF patient stratification. parameters with immune cell parameters have stronger associations with immunotherapy outcome. A predictive signature classifier computed from all available tissue data Conclusions is also under consideration. This study examines the clinical significance of The use of two independent image analysis tech- cancer-associated fibroblast (Thy1, SMA, FAP) pro- nologies to assess biomarkers and the concordant files in pretreatment tumor specimens to determine results from cell counts and QIF adds confidence in their association with immunotherapy outcome in thefindings. TheAQUA methodofQIF measures melanoma. We find that FAP, by both digital cell protein expression as cumulative signal intensity per counts and quantitative immunofluorescence of pro- unit compartment area, and it has been shown to be tein expression, shows significant positive associa- proportional to analyte concentration [35]. This is tions with survival outcome. The positive association fundamentally different from counts of digitally phe- is independent of baseline variables in multivariable notyped cells [36]. The similar results of the two analyses. In contrast, FAP is inversely associated with methodologies suggest shared biological relevance. prognosis in the absence of immunotherapy in a his- However, cell counts use intuitive absolute units torical cohort. The novel discovery that FAP is a and exhibited stronger associations with survival negative prognostic and positive indicative biomarker outcome than QIF, therefore, it may have a greater in melanoma suggests mechanistic involvement in potential for clinical translation in digital precision anti-PD-1 survival advantage. Its independence from immuno-oncology. previously described biomarkers like CD8 and PD-L1 In summary, this study demonstrates that pretreat- suggestitcould have valueincombinationwith ment CAF profiles, by in situ cell counts or QIF those markers to more accurately predict outcome to protein expression, are independently associated with immunotherapy. Table 3 Univariable and multivariable Cox regression analyses for overall survival of melanoma patients and CAF parameters by cell counts Variable Control OS Anti-PD-1 OS (LO/HI) a a Univariable analysis Univariable analysis Multivariable analysis per variable Multivariable analysis with Thy1, SMA, FAP HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value Thy1 /total 1.65 (0.87–2.88) 0.12 4.66 (2.34–8.82) < 0.0001 4.67 (2.19–9.53) 0.0001 3.02 (1.44–6.10) 0.0044 + + SMA /Thy1 0.45 (0.22–1.07) 0.070 0.34 (0.18–0.68) 0.0029 0.32 (0.16–0.67) 0.0027 0.62 (0.30–1.31) 0.20 + + FAP /Thy1 0.57 (0.37–0.88) 0.012 4.11 (2.05–9.14) < 0.0001 4.64 (2.27–10.52) < 0.0001 3.61 (1.65–8.56) 0.0011 Abbreviations: CAF cancer-associated fibroblast, CI confidence interval, HI high, HR hazard ratio, LO low, OS overall survival Cox proportional hazards model included age, sex, mutation status, stage, treatment, and prior immune checkpoint blockade as covariates Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 9 of 10 Additional file Received: 21 January 2019 Accepted: 11 July 2019 Additional file 1: Supplementary Figure 1. Linear regressions of CAF parameters in melanoma by cell counts and quantitative References immunofluorescence. Correlation between cell counts and QIF scores for 1. Topalian SL, Hodi FS, Brahmer JR, et al. Safety, activity, and immune CAF (Thy1, SMA, FAP) markers (A). Relationships between Thy1, SMA, and correlates of anti-PD-1 antibody in cancer. N Engl J Med. 2012;366(26): FAP by cell counts and QIF (B). Abbreviations: AU, arbitrary units; CAF, 2443–54. cancerassociated fibroblast; QIF, quantitative immunofluorescence. (PDF 2. Weber JS, D'Angelo SP, Minor D, et al. Nivolumab versus chemotherapy in 439 kb) patients with advanced melanoma who progressed after anti-CTLA-4 treatment (CheckMate 037): a randomised, controlled, open-label, phase 3 trial. Lancet Oncol. 2015;16(4):375–84. Acknowledgements 3. Wolchok JD, Chiarion-Sileni V, Gonzalez R, et al. Overall survival with The authors thank Lori A. Charette and the staff of Yale Pathology Tissue combined Nivolumab and Ipilimumab in advanced melanoma. N Engl J Services for expert histology services. This work is based on the PhD Med. 2017;377(14):1345–56. dissertation research of Dr. Pok Fai Wong as a Gruber Science Fellow at 4. Robert C, Schachter J, Long GV, et al. Pembrolizumab versus Ipilimumab in Yale University. advanced melanoma. N Engl J Med. 2015;372(26):2521–32. 5. Ribas A, Hamid O, Daud A, et al. Association of Pembrolizumab with Tumor Response and Survival among Patients with Advanced Melanoma. JAMA. Authors’ contributions 2016;315(15):1600–9. Conception and design: PFW, DLR. Case selection and clinical data collection: 6. Eggermont AMM, Blank CU, Mandala M, et al. Adjuvant Pembrolizumab PFW, JWS, HMK, DLR. Multiplex cancer-associated fibroblast analysis: PFW. versus placebo in resected stage III melanoma. N Engl J Med. 2018;378(19): Multiplex immune marker analysis: PFW, SG. Statistical analysis: PFW, WW, DZ. 1789–801. Drafting of manuscript: PFW, DLR. Critical revision of manuscript: All authors. 7. Abdel-Wahab N, Shah M, Suarez-Almazor ME. Adverse events associated Final approval of manuscript: All authors. Study supervision: DLR. with immune checkpoint blockade in patients with Cancer: a systematic review of case reports. PLoS One. 2016;11(7):e0160221. 8. Tirosh I, Izar B, Prakadan SM, et al. Dissecting the multicellular ecosystem of Funding metastatic melanoma by single-cell RNA-seq. Science. 2016;352(6282):189–96. This work was supported by funds from Navigate BioPharma (Novartis subsidiary), 9. Lambrechts D, Wauters E, Boeckx B, et al. Phenotype molding of stromal Yale SPORE in Lung Cancer and Yale Cancer Center to D.L. Rimm; R01 CA227473, cells in the lung tumor microenvironment. Nat Med. 2018;24(8):1277–89. K24CA172123 and P50 CA121974 to H.M. Kluger; and the Gruber Science 10. Fiore VF, Strane PW, Bryksin AV, White ES, Hagood JS, Barker TH. Conformational Fellowship to P.F. Wong from the Gruber Foundation. The funding coupling of integrin and Thy-1 regulates Fyn priming and fibroblast sources had no role in study design; collection, analysis and interpretation of mechanotransduction. J Cell Biol. 2015;211(1):173–90. data; preparation of the manuscript or the decision to submit for publication. 11. Hinz B, Celetta G, Tomasek JJ, Gabbiani G, Chaponnier C. Alpha-smooth muscle actin expression upregulates fibroblast contractile activity. Mol Biol Cell. 2001;12(9):2730–41. Availability of data and materials 12. Wong PF, Gall MG, Bachovchin WW, McCaughan GW, Keane FM, Gorrell MD. De-identified datasets used and/or analyzed during the current study are Neuropeptide Y is a physiological substrate of fibroblast activation protein: available from the corresponding author upon reasonable request. enzyme kinetics in blood plasma and expression of Y2R and Y5R in human liver cirrhosis and hepatocellular carcinoma. Peptides. 2016;75:80–95. 13. Garin-Chesa P, Old LJ, Rettig WJ. Cell surface glycoprotein of reactive Ethics approval and consent to participate stromal fibroblasts as a potential antibody target in human epithelial The study was approved by the Yale Human Investigation Committee protocol cancers. Proc Natl Acad Sci U S A. 1990;87(18):7235–9. #9505008219 and conducted in accordance with the Declaration of Helsinki. All 14. Feig C, Jones JO, Kraman M, et al. Targeting CXCL12 from FAP-expressing patients provided written informed consent or waiver of consent in carcinoma-associated fibroblasts synergizes with anti-PD-L1 immunotherapy circumstances stipulated by the HIC protocol. in pancreatic cancer. Proc Natl Acad Sci U S A. 2013;110(50):20212–7. 15. Ozdemir BC, Pentcheva-Hoang T, Carstens JL, et al. Depletion of carcinoma- associated fibroblasts and fibrosis induces immunosuppression and Consent for publication accelerates pancreas cancer with reduced survival. Cancer Cell. 2014; Not applicable. 25(6):719–34. 16. Kumar V, Donthireddy L, Marvel D, et al. Cancer-associated fibroblasts neutralize the anti-tumor effect of CSF1 receptor blockade by inducing PMN-MDSC infiltration of tumors. Cancer Cell. 2017;32(5):654–668.e655. Competing interests 17. Wong PF, Wei W, Smithy JW, et al. Multiplex quantitative analysis of tumor- D.L. Rimm has served as a consultant, advisor or served on a Scientific infiltrating lymphocytes and immunotherapy outcome in metastatic Advisory Board for Amgen, Astra Zeneca, Agendia, Biocept, Bristol-Meyers melanoma. Clin Cancer Res. 2019;25(8):2442-49. Squibb, Cell Signaling Technology, Cepheid, Daiichi Sankyo, GSK, Merck, 18. Qin Y, Petaccia de Macedo M, Reuben A, et al. Parallel profiling of immune NanoString, PerkinElmer, PAIGE, and Ultivue. He has received research fund- infiltrate subsets in uveal melanoma versus cutaneous melanoma unveils ing from Astra Zeneca, Cepheid, Navigate/Novartis, NextCure, Lilly, Ultivue, similarities and differences: a pilot study. Oncoimmunology. 2017;6(6): and PerkinElmer. H.M. Kluger has served as a consultant for Corvus, Nektar, e1321187. Biodesix, Genentech, Pfizer, Merck and Celldex, and has received research support from Merck, Apexigen and Bristol-Meyers Squibb. All other authors 19. Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria have no competing interests. in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer. 2009; 45(2):228–47. Author details 20. Brown JR, Wimberly H, Lannin DR, Nixon C, Rimm DL, Bossuyt V. Department of Pathology, Yale School of Medicine, New Haven, CT 06510, Multiplexed quantitative analysis of CD3, CD8, and CD20 predicts USA. Yale Cancer Center, Yale School of Medicine, New Haven, CT 06510, response to neoadjuvant chemotherapy in breast cancer. Clin Cancer USA. Department of Biostatistics, Yale School of Public Health, New Haven, Res. 2014;20(23):5995–6005. CT 06510, USA. Section of Medical Oncology, Department of Internal 21. Huang W, Hennrick K,DrewS.Acolorful future of quantitative Medicine, Yale School of Medicine, New Haven, CT 06510, USA. Department pathology: validation of Vectra technology using chromogenic of Pathology, Yale School of Medicine, 310 Cedar St, BML 116, PO Box multiplexed immunohistochemistry and prostate tissue microarrays. 208023, New Haven, CT 06520, USA. Hum Pathol. 2013;44(1):29–38. Wong et al. Journal for ImmunoTherapy of Cancer (2019) 7:194 Page 10 of 10 22. Camp RL, Chung GG, Rimm DL. Automated subcellular localization and quantification of protein expression in tissue microarrays. Nat Med. 2002; 8(11):1323–7. 23. Lausen B, Schumacher M. Maximally selected rank statistics. Biometrics. 1992;48(1):73–85. 24. Eton O, Legha SS, Moon TE, et al. Prognostic factors for survival of patients treated systemically for disseminated melanoma. J Clin Oncol. 1998;16(3): 1103–11. 25. Manola J, Atkins M, Ibrahim J, Kirkwood J. Prognostic factors in metastatic melanoma: a pooled analysis of eastern cooperative oncology group trials. J Clin Oncol. 2000;18(22):3782–93. 26. Neuman HB, Patel A, Ishill N, et al. A single-institution validation of the AJCC staging system for stage IV melanoma. Ann Surg Oncol. 2008;15(7):2034–41. 27. Joosse A, Collette S, Suciu S, et al. Sex is an independent prognostic indicator for survival and relapse/progression-free survival in metastasized stage III to IV melanoma: a pooled analysis of five European organisation for research and treatment of cancer randomized controlled trials. J Clin Oncol. 2013;31(18): 2337–46. 28. Hartman DJ, Ahmad F, Ferris RL, Rimm DL, Pantanowitz L. Utility of CD8 score by automated quantitative image analysis in head and neck squamous cell carcinoma. Oral Oncol. 2018;86:278–87. 29. Hugo W, Zaretsky JM, Sun L, et al. Genomic and transcriptomic features of response to anti-PD-1 therapy in metastatic melanoma. Cell. 2016;165(1):35–44. 30. Zhao F, Evans K, Xiao C, et al. Stromal fibroblasts mediate anti-PD-1 resistance via MMP-9 and dictate TGFbeta inhibitor sequencing in melanoma. Cancer Immunol Res. 2018;6(12):1459–71. 31. Lakins MA, Ghorani E, Munir H, Martins CP, Shields JD. Cancer-associated fibroblasts induce antigen-specific deletion of CD8+ T cells to protect tumour cells. Nat Commun. 2018;9(1):948. 32. Cancer Genome Atlas Network. Genomic classification of cutaneous melanoma. Cell. 2015;161(7):1681–96. 33. Tumeh PC, Harview CL, Yearley JH, et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature. 2014;515(7528):568–71. 34. Chen PL, Roh W, Reuben A, et al. Analysis of immune signatures in longitudinal tumor samples yields insight into biomarkers of response and mechanisms of resistance to immune checkpoint blockade. Cancer Discov. 2016;6(8):827–37. 35. Toki MI, Cecchi F, Hembrough T, Syrigos KN, Rimm DL. Proof of the quantitative potential of immunofluorescence by mass spectrometry. Lab Investig. 2017;97(3):329–34. 36. Carvajal-Hausdorf DE, Schalper KA, Neumeister VM, Rimm DL. Quantitative measurement of cancer tissue biomarkers in the lab and in the clinic. Lab Investig. 2015;95(4):385–96. Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Journal

Journal for ImmunoTherapy of CancerSpringer Journals

Published: Jul 23, 2019

There are no references for this article.