Characterization of Immune Regulatory Molecules B7-H4 and PD-L1 in Low and High Grade Endometrial Tumors

YGYNO-976673; No. of pages: 7; 4C: 3, 4, 5, 6 Gynecologic Oncology xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Gynecologic Oncology

journal homepage: www.elsevier.com/locate/ygyno

Society position statements/white papers Characterization of immune regulatory molecules B7-H4 and PD-L1 in low and high grade endometrial tumors

Amy Bregar a,b,1,AmitDeshpanded,1,ChrisGranged,TongZid,JenniferStallc, Heather Hirsch d,JasonReevesd, Sriram Sathyanarayanan d, Whitﬁeld B. Growdon a,b,BoR.Ruedaa,b,⁎ a Vincent Center for Reproductive Biology, Vincent Department of Obstetrics and Gynecology, Massachusetts General Hospital, Boston, MA 02114, United States b Gynecologic Oncology Division, Vincent Department of Obstetrics & Gynecology, Massachusetts General Hospital, Boston, MA 02114, United States c Department of Pathology, Massachusetts General Hospital, Boston, MA 02114, United States d Jounce Therapeutics, Inc., Cambridge, MA, United States

HIGHLIGHTS

• Endometrial tumors with microsatellite instability and high grade have elevated PD-L1 protein levels. • Protein levels of B7-H4 are prevalent and independent of microsatellite instability, grade and histology. • Combination immunotherapies directed against checkpoint proteins may hold promise for endometrial carcinoma.

article info abstract

Article history: Background. The objective of this investigation was to characterize the expression landscape of immune reg- Received 31 December 2016 ulatory molecules programmed death-ligand-1 (PD-L1, B7-H1) and B7-H4 in a cohort of endometrial tumors Received in revised form 6 March 2017 across the spectrum of grade and histology. Accepted 10 March 2017 Materials and methods. With institutional review board approval, 70 endometrial tumors from patients with Available online xxxx known clinical outcomes were identified representing a spectrum of grade and histology. Immunohistochemis- try (IHC) was performed for PD-L1 and B7-H4 and scored. Microsatellite instability (MSI) status was assessed for Keywords: Endometrial carcinoma endometrioid tumors using the institutional IHC assay for expression of the mismatch repair (MMR) genes, Immune checkpoint inhibition MLH1, MSH2, MSH6 and PMS2. RNA sequencing data from the Cancer Genome Atlas was queried for expression PD-L1 over-expression levels of CD274 (PD-L1 protein) and VTCN1 (B7-H4) across molecular subtypes of endometrial carcinoma and B7-H4 expression were correlated with a T cell infiltration index. Results. We identified 40 low grade endometrioid tumors and a cohort of 30 high grade tumors. PD-L1 expression was observed in both high and low grade endometrial tumors (56% vs 35%, p = 0.07). In the low grade tumors, PD-L1 expression was associated with MSI status (p b 0.01). The high grade cohort had similar rates of PD- L1 expression compared to low grade MSI tumor (56% and 62% respectively), and both were distinct from low grade MSS tumors (22%, p b 0.05). High (3+) B7-H4 positive cells were observed in both high and low grade carcinomas (33% and 31% respectively). RNA profiling data from confirmed highest CD274 expression in POLE and MSI tumors that was linearly correlated with T cell infiltration, while VTCN1 expression appeared consistent across molecular subtypes. Conclusions. While PD-L1 expression correlated with MSI and high grade tumors, B7-H4 expression was independent of grade, histology and immune cell infiltration. The development and testing of multi-agent therapeutics targeting PD-L1 and B7-H4 may be a novel strategy for endometrial tumors. © 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction ⁎ Corresponding author at: Massachusetts General Hospital, Vincent Center for Reproductive Biology, Vincent Department of Obstetrics and Gynecology, THR 933, 55 Endometrial cancer (EnCa) is the most common gynecologic malig- Fruit Street, Boston, MA 02114, United States. E-mail address: [email protected] (B.R. Rueda). nancy accounting for over 60,000 newly diagnosed cases and over 8000 1 Equal contribution. deaths in the United States in 2016 [1]. Marked differences in clinical

http://dx.doi.org/10.1016/j.ygyno.2017.03.006 0090-8258/© 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article as: A. Bregar, et al., Characterization of immune regulatory molecules B7-H4 and PD-L1 in low and high grade endometrial tumors, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.03.006 2 A. Bregar et al. / Gynecologic Oncology xxx (2017) xxx–xxx behavior have been observed in patients with EnCa depending on the Expression of B7-H4 has also been reported to inversely correlate with histologic subtype, the tumor grade and the extent of cancer spread. In- Tcellinfiltration [22]. vestigators have suggested a classification system that separates endo- EnCa is a tumor marked by diverse grade and histology with signif- metrial tumors into type I and type II subsets to account for the icant subsets harboring high mutational burdens in concert with MSI striking divide in risk factors, clinical behavior and approach to therapy [4]. We hypothesize that this intrinsic nature predisposes EnCa to [2]. While type II cancers account for only 15–25% of all EnCa, patients being responsive to this new class of immune sensitization therapy with these tumors account for 75% of the mortality observed highlight- and that understanding the baseline expression profiles of key immune ing the need for novel therapies beyond conventional surgery, radiation checkpoint proteins in endometrial tumors will be crucial for identify- and cytotoxic chemotherapy for this subset of tumors [3]. ing those patients most likely to respond this new class of immunother- Recently, the clinical type I/II distinction has been reframed into mo- apies. We analyzed the TCGA data for expression of PD-L1 and B7-H4 lecular categories by TCGA with distinctive molecular signatures that and found that immune infiltration signatures as well as PD-L1 and define prognosis [4]. The categories described are 1) POLE associated B7-H4 expression were manifest most in MSI and POLE tumors when with mutation in the DNA polymerase gene E marked by ultramutation, compared to MSS tumors. We then conducted a molecular analysis of 2) microsatellite instability (MSI) associated with mutations in the DNA 70 endometrial tumors from the range of histologic subtypes and repair machinery, marked by hypermutation, 3) microsatellites stable found that high grade tumors and those low grade endometrioid tumors (MSS) marked by endometrioid histology and 4) DNA copy number that harbor MSI express higher PD-L1 and B7-H4 protein levels when high marked by serous-like histology. Importantly, copy number high, compared to low grade endometrioid MSS tumors. Collectively, these predominantly serous like tumors portended the most guarded progno- data suggest that a range of endometrial tumors harbor signatures sis, with the ultramutated POLE group being associated with the best that may be amenable to immune modulation through interactions prognosis [4]. with PD-1 and B7-H4. Endometrial cancers with POLE mutation or MSI tumors can harbor 10 to 100 times more mutations as compared to the MSS tumors [4]. 2. Materials and methods Those tumors with high mutational burden can harbor potent antigens and therefore are subject to host immune surveillance which is possibly 2.1. Patients and samples associated with improved prognosis observed in these tumors [5–7]. Given this hypothesis, when these subsets of tumors do recur, there is Under an institutionally approved research plan and agreement, we increased reliance on immune cloaking mechanisms for survival and identified 70 endometrial carcinomas across a spectrum of grade and metastasis [8]. These tumors may contain prominent immune cell infil- histology of patients diagnosed between 2010 and 2013 with complete trates and express high levels of immune checkpoint molecules that clinical follow up and available tissue for analysis. Clinical factors were dampen the host immune response resulting in tumor growth [9]. extracted from patient records, including age, grade, stage, treatment, Recent clinical studies have demonstrated durable clinical benefit recurrence and survival. All molecular analyses were carried out on for- with antibody therapies that target immune checkpoints [9–11]. Pro- malin-fixed and paraffin-embedded (FFPE) diagnostic specimens. He- grammed death 1 (PD-1, B7-H1) is a member of B7 receptors that mod- matoxylin and eosin-stained slides were marked for tumor location by ulate T cell response [12]. Expression of the checkpoint molecule PD-1 a gynecologic oncology pathologist. on immune lymphocytes has been shown to limit T cell response through the interaction with either or both of its ligands, PD-L1 and 2.2. Immunohistochemistry PD-L2 [12]. Indeed, as a mechanism of immune evasion, tumors have been shown to upregulate PD-L1 expression on tumor cells further sup- Paraffin embedded EnCa FFPE tissue sections of 5 μm thickness were pressing T cell activity promoting immune evasion and enhancing subjected to immunohistochemistry (IHC) for PD-L1 and B7-H4 using tumor survival [13]. Antibody therapies that target PD-1 clones D1M8I (B7-H4), E1L3N (PD-L1) (Cell Signaling, Waltham MA), (Pembrolizumab, Nivolumab, etc.) or PD-L1 (Atezolizumab, Avelumab, following protocols established for staining on the Leica BONDRX auto- etc.), and disrupt the PD-1/PD-L1 interaction, have shown marked mated platform. Scoring for both markers was performed independent- anti-tumor efficacy in numerous indications including, melanoma, ly by two operators according to guidelines. Briefly, B7-H4 expression non–small-cell lung cancer, renal-cell carcinoma, and Hodgkin's lym- was scored on an intensity scale of 0 to 3+ and reported as an H- phoma [10]. These immune checkpoint blockade therapies neutralize score calculated as (intensity score) ∗ (% cells for each staining the inhibitory signals generated the checkpoint molecules resulting in induction, activation and expansion of T cells that eliminate the tumor Table 1 [13]. Cohort Characteristics. Recent clinical studies suggest that MSI status predicted clinical ben- Age (range) 64.9 (39.6–88.7) efit to anti-PD-1 therapy with Pembrolizumab [14]. In this study, Stage 1 50 (71%) patients with tumors harboring MSS had an 11% clinical benefitrate 2 3 (4%) (CR + PR + SD) compared to a 70% clinical benefit rate in patients 3 9 (13%) 4 8 (11%) with tumors with MSI [14]. The MSI tumors presented with an over Grade 1 37 (53%) ten-fold higher mutational burden and significantly elevated levels of 2 4 (6%) PD-L1 expression at the invasive front of the tumors when compared 3 29 (41%) to the MSS tumors [14]. Numerous other correlative studies in addition- Histology High grade endometrioid 10 (14%) Low grade endometrioid 40 (58%) al indications have echoed these results and found significantly higher Carcinosarcoma 10 (14%) response rates in those tumors expressing PD-L1 [9,15]. Uterine serous carcinoma 10 (14%) Similar to PD-L1, another member of the B7 family, B7-H4 (a.k.a. MSI Status Endometrioid MSI-high 13 (33%) VTCN1, B7h.5, B7S1, B7x) suppresses effector function of T cells through Endometrioid MSS 27 (67%) interaction with an unknown receptor expressed on T cells. Although Not assessed 30 Recurrence (%) Total 15 (21%) the exact mechanism of T cell inhibition is not known, based on pub- High grade endometrioid 1 (10%) lished reports B7-H4 is thought to mediate tumor immune evasion by Carcinosarcoma 5 (50%) inhibiting activation of T and NK cells [16,17]. B7-H4 has been reported Low grade endometrioid MSI-high 1 (8%) to be frequently overexpressed by tumor cells in a variety of solid tu- Low grade endometrioid MSS 3 (13%) Uterine serous carcinoma 5 (50%) mors including ovarian, lung, breast, pancreatic and renal cell [16–21].

Fig. 1. Immunohistochemistry for clinical detection of microsatellite instability. These panels depict representative images of the protein levels detected by immunohistochemistry for the mismatch repair proteins MLH1 (a), PMS2 (b), MSH2 (c) and MSH6 (d) which were assessed on all low grade endometrioid endometrial carcinomas in our cohort (n = 40). No specimens were found to harbor loss of MSH2 or MSH6, however 13 tumors in the cohort had simultaneous loss of expression of MLH1 and PMS2 (Panel A) and were therefore classified as MSI high. All images were 10× magnification. intensity). In addition to the H-score, tumors were classified based on 2.3. Bioinformatics the Gestalt score into B7-H4 low (0 or 1+ score) or high (2+ or 3+). Four of the 10 carcinosarcoma samples had a non-diagnostic signal for In addition to IHC, expression of CD274 and VTCN1 (genes coding B7-H4 and were excluded from analyses. PD-L1 expression was scored PD-L1 and B7-H4 respectively) was evaluated using publically available as % PD-L1 positive cells. Cells were counted positive regardless of ex- RNA sequencing data collected from TCGA. Raw sequence data was pre- pression on tumor or infiltrating immune cells. Two different cutoffs pared and gene expression values utilizing the fragments per kilobase of (≥1% and ≥5%) were evaluated for PD-L1. Microsatellite instability exon per million fragments mapped (FPKM) were computed by (MSI) status was assessed for the endometrioid tumors (n = 40) OmicSoft Corporation utilizing their proprietary data normalization using the institutional clinical IHC assay for expression of mismatch re- and expression analysis pipeline. All the FPKM values were converted pair protein expression of MLH1, MSH2, MSH6 and PMS2. Each sample using a logarithmic function with a base 2 and analyzed. Further clinical was reviewed and scored by a gynecologic oncology specialist patholo- attributes and subgroups based on gene expression profiling and muta- gist. Samples lacking positive signal for MLH1 and PMS2 expression tional analysis performed by the TCGA were also obtained from were classified as MSI positive. OmicSoft for use in this analysis. Gene expression levels of CD274 and

Table 2 PD-L1 and B7-H4 expression proﬁles.

Criteria A (1%) Criteria B (5%)

PD-L1 negative (%) PD-L1 positive (%) p-value PD-L1 negative PD-L1 positive p-value

Histology subtypes Low grade endometrioid MSS 21 (78%) 6 (22%) referent 25 (93%) 2 (7%) referent Low grade endometrioid MSI-high 5 (38%) 8 (62%) 0.03 7 (54%) 6 (46%) b0.005 Uterine serous carcinoma 7 (70%) 3 (30%) 0.24 8 (80%) 2 (20%) 0.88 High grade endometrioid 3 (30%) 7 (70%) 0.02 5 (50%) 5 (50%) 0.1 Carcinosarcoma 3 (30%) 7 (70%) 0.02 6 (60%) 4 (40%) 0.28 Grade subtypes Low grade endometrial carcinoma 26 (65%) 14 (35%) referent 32 (80%) 8 (20%) referent High grade endometrial carcinoma 13 (44) 17 (56%) 0.05 19 (63%) 11 (37%) 0.12 Low grade endometrioid MSI-high 5 (38%) 8 (62%) referent 7 (54%) 6 (46%) referent High grade endometrial carcinoma 13 (44) 17 (56%) 0.77 19 (63%) 11 (37%) 0.56

Criteria A Criteria B

B7-H4 low expression (%) B7-H4 high expression (%) p-value Average H-score p-value

Histology subtypes Endometrioid MSS 16 (59%) 11 (41%) referent 107.2 referent Endometrioid MSI-high 3 (23%) 10 (77%) 0.06 170 0.07 Uterine serous carcinoma 3 (30%) 7 (70%) 0.15 136.5 0.45 High grade endometrioid 2 (20%) 8 (80%) 0.06 172.5 0.1 Carcinosarcomaa 1 (17%) 5 (83%) 0.09 135.8 0.55 Grade subtypes Low grade endometrial carcinoma 19 (48%) 21 (52%) referent 127 referent High grade endometrial carcinomaa 6 (23%) 20 (77%) 0.07 149 0.41 Low grade endometrioid MSI-high 3 (23%) 10 (77%) referent 170 referent High grade endometrial carcinomaa 6 (23%) 20 (77%) 0.99 149 0.59

a 4 samples were non-diagnostic.

VTCN1 were plotted alongside the described molecular classifications, 30 high grade tumors including 10 grade 3 endometrioid tumors, 10 grade, histologic subtype and a Th1 signature that was used a surrogate uterine serous carcinomas and 10 carcinosarcomas. At five years, 90% for T cell activation and tumor infiltration. of the subjects were still alive. No correlation was observed between expression of PD-L1 and B7-H4 and age, stage or recurrence risk. 2.4. Statistical analysis 3.2. MSI Status Two-sided Fisher's exact tests and χ2 were utilized to compare pro- portions for univariate analysis. Correlation was assessed utilizing Pear- The low grade endometrial tumors were subjected to protein analy- son correlation and Spearman rank sum testing as appropriate. sis of MMR proteins. 13 (33%) cases were identified as defective for Continuous variables were compared utilizing t-tests and linear regres- MMR (dMMR), and therefore MSI-positive, using the criteria described sion models. Kaplan–Meier survival estimates were generated from in Materials and methods. None of the tumors in this cohort exhibited date of histological diagnosis to time of last follow up or death. Log- loss of MSH2 or MSH6. Fig. 1 shows representative images of MSI and rank tests were utilized to determine statistical significance of survival MSS IHC staining. curves. An α b 0.05 defined statistical significance. Analysis was performed on STATA version 10.0 (College Station, TX). 3.3. PD-L1 and B7-H4 expression profiles

3. Results PD-L1 expression was assessed in all 70 endometrial cancer tumors utilizing two scoring criteria, a) a low threshold for PD-L1 expression 3.1. Clinical characteristics ≥1% of total cells (criteria A) and b) a high threshold ≥5% of total cells (criteria B) positive. The results from the analysis are summarized in We identiﬁed 70 endometrial carcinomas representing a full spec- Table 2. Representative sections of each score are depicted in Fig. 2A. trum of grade, stage and histology (Table 1). The average age of the co- PD-L1 positivity was observed in a small subset of low grade MSS tu- hort was 67 with 75% of cases representing stage I disease. The cohort mors 22% (criteria A) or 7% (criteria B). In contrast, PD-L1 expression incorporated 40 low grade (grade 1 and 2) endometrioid tumors, and was signiﬁcantly higher in the MSI patients, 62% (criteria A) or 46%

Fig. 2. (A) PD-L1 Scoring Criteria. These panels represent 3 individual tumor samples and demonstrate the range of PD-L1 protein levels detected utilizing two criteria to assign positivity (criteria A and B). Criteria A utilized a cutoff level of positive staining on N1% of carcinoma cells. Criteria B is more stringent and utilized a cutoff level of positive staining on N5% of carcinoma cells. The upper panels are 10× and the lower panels are 1× magniﬁcation (B) Correlation of expression with MSI status showed a signiﬁcantly greater presence of PD-L1+ cells in MSI-high subsets of endometrial tumors utilizing both criteria A (p b 0.03) and criteria B (p b 0.005).

Fig. 3. B7-H4 Scoring Criteria. These panels represent 4 individual tumor samples and demonstrate the range of B7-H4 protein levels detected. An intensity score and a % carcinoma stained proportion was assigned for each sample and an H-Score was generated by multiplying the two indexes. The upper panels are 5× magniﬁcation while the lower panels are 1 x magniﬁcation.

(criteria B) positive staining observed in the MSI cohort (p b 0.03 and p B7-H4 modeled as a continuous variable revealed a trend towards ele- b 0.005 respectively, Fig. 2B; Table 2). In addition, PD-L1 expression was vated expression with increasing grade, though this was not statistically observed in the high grade carcinomas with 56% (criteria A) and 37% significant (p = 0.07). Importantly, there was no significant difference (criteria B) positive expression, elevated when compared to the low in B7-H4 expression between the MSI and MSS tumors (p = 0.07; grade endometrioid cohort as a group although this was not statistically Table 2). Logistic regression correlating protein levels of B7-H4 with significant (p = 0.05; p = 0.12 by criteria A and criteria B respectively). PD-L1 (criteria A) of the entire cohort with both levels (n = 66) re- B7-H4 expression was evaluated in this set of endometrial tumors vealed a significant positive association (p b 0.02). Unlike the low using an IHC assay. Representative images and H-score criteria are pre- grade cohort where co-expression of PD-L1 and B7-H4 affected 22% of sented in Fig. 3A. B7-H4 expression was observed in both high grade the cohort, the high grade cohort exhibited a 50% co-expression rate and low grade endometrial tumors (Fig. 3BandTable 2). Expression of (p b 0.02) (Fig. 4).

Fig. 4. Co-expression of PD-L1 (criteria A) and B7-H4 across EnCa histologic subtypes. In contrast to low grade endometrial carcinomas, high grade endometrial carcinomas harbored a 50% co-expression of both PD-L1 (criteria A) and B7-H4 (p b 0.01). This effect was primarily due to high grade endometrioid and carcinosarcoma subtypes and was statistically identical to the low grade endometrial subset with MSI. Notably, the uterine serous carcinomas demonstrated a low level of co-expression.

3.4. RNA expression profiling of CD274 and VTCN1 across molecular endo- biomarker for predicting response to anti-PD-1 or anti-PD-L1 therapies metrial carcinoma subtypes and an IHC assay to detect PD-L1 expression has been approved by the FDA as a diagnostic assay to identify patients with a greater likelihood Expression of CD274 and VTCN1 mRNA were analyzed in a cohort of of responding to Pembrolizumab therapy in patients with Non-Small 383 endometrial tumor samples available as a part of the TCGA dataset Cell Lung Cancer (NSCLC). and quantitated as either maximum, average, or minimum expression. Limited clinical data exists investigating the use of checkpoint inhi- Heat map analysis revealed that CD274 was highly expressed in the bition in endometrial cancer [23,24]. The growing recognition that MSI POLE and MSI and subsets with the lowest expression noted in the may serve as a surrogate biomarker for response to immunotherapy copy number low. CD274 expression exhibited significant correlation has resulted in women with advanced endometrial cancer receiving (r2 = 0.564) with an established Th1 signature to approximate lym- anti-PD-1 therapy [5]. This study is focused on evaluating expression phocyte infiltration when modeled as quartiles and continuously. Im- of PD-L1 and B7-H4, as well as the MSI status in a cohort of endometrial portantly, maximum VTCN1 expression was observed in all subsets tumors. Using IHC staining and analysis of RNA expression dataset from with no clear over-representation in ultra or hyper mutated subsets. TCGA, significant portions of endometrial tumors were observed to ex- No correlation of VTCN1 expression and Th1 signature was observed press high levels of PD-L1 and B7-H4. While B7-H4 levels were not sta- (r2 = 0.004). These data are summarized in Fig. 5. tistically different across various endometrial carcinoma subsets tested, the high prevalence of PD-L1 positivity was observed in low grade 4. Discussion endometrioid MSI-high cohort. This data is consistent with high levels of PD-L1 expression in MSI-high colorectal cancers [14]. Analysis of Sensitizing the host immune system to solid tumors through the use RNA expression from the TCGA dataset confirmed these findings and of immunotherapies holds the promise of therapeutic benefitacrossdis- further correlated expression of PD-L1 with a lymphocyte infiltration ease sites. In melanoma and NSCLC, while significantly higher response index. Comparisons of normal, hyperplastic and malignant endometri- to immune checkpoint modulators is observed in tumors with high PD- um have demonstrated increased B7-H4 levels further suggesting endo- L1 positivity, a small subset of patients with low PD-L1 expression also metrial tumors may rely on immune cloaking to persist [25]. benefited from anti-PD-1 or PD-L1 therapies [10,15]. Additional bio- Importantly, we observed that the high grade EnCa tumors were more markers, including genomic instability resulting in high mutational likely to present with simultaneously elevated PD-L1 and B7-H4 levels load or strong oncogenic events including viral oncogenes could con- when compared to the low grade cohorts, irrespective of the MSI status. tribute to sensitivity to anti-PD-1 therapies. These patients are more Collectively, these data suggest that immune modulation is an impor- likely to respond to immune checkpoint therapies due to their increased tant mechanism for significant subsets of EnCa, one that may be antigenicity. PD-L1 expression is currently the most prevalent exploited through therapeutic inhibition of PD-L1 and possibly other

Fig. 5. RNA expression profiling analysis of CD274 and VTCN1 from the Cancer Genome Atlas (TCGA). Panel A depicts the 4 molecular subgroups defined in the TCGA investigation of 383 endometrial carcinomas and demonstrates that the POLE and MSI subsets have the greatest proportion of lymphocyte infiltration based on a validated Th1 signature that is utilized as a surrogate marker. We observed higher levels of expression of the CD274, the gene coding PD-L1, in the Th1 top 25% quartile in each of the molecular subgroups. In addition to T cell infiltration, the greatest mutational burden was observed in the POLE and MSI subsets of endometrial tumors. Panel B examines the POLE and MSI subsets of tumors and plots the Th1 signature against expression of CD274 and VTCN1 finding a significant linear correlation with CD274 and Th1 index expression (r2 = 0.56), while there was consistent expression of VTCN1 across various levels of Th1 index expression (r2 = 0.004). These data suggest the message for PD-L1 is more dependent upon histologic tumor variant anddegreeof infiltrating lymphocytes, while the message for B7-H4 is not associated with T cell infiltration.

B7 family immune checkpoint proteins. These preliminary data demon- [4] Cancer Genome Atlas Research N, C. Kandoth, N. Schultz, A.D. Cherniack, R. Akbani, Y. Liu, et al., Integrated genomic characterization of endometrial carcinoma, Nature strate an immune landscape that provides a strong rationale for use of 497 (7447) (2013) 67–73. PD-1/PD-L1 inhibition in EnCa with MSI and high grade. [5] P. Gargiulo, C. Della Pepa, S. Berardi, D. Califano, S. Scala, L. Buonaguro, et al., Tumor Published data specifically related to checkpoint inhibition in EnCa genotype and immune microenvironment in POLE-ultramutated and MSI- hypermutated endometrial cancers: new candidates for checkpoint blockade im- are limited to retrospective analyses of women with EnCa who have munotherapy? Cancer Treat. Rev. 48 (2016) 61–68. demonstrated an exceptional response to anti-PD-1 treatment. Analy- [6] A. Snyder, V. Makarov, T. Merghoub, J. Yuan, J.M. Zaretsky, A. Desrichard, et al., Ge- ses of these tumors have demonstrated that they uniformly harbored netic basis for clinical response to CTLA-4 blockade in melanoma, N. Engl. J. Med. – a high mutational load, due to either MSI or POLE mutations further 371 (23) (2014) 2189 2199. [7] L.M. Colli, M.J. Machiela, T.A. Myers, L. Jessop, K. Yu, S.J. Chanock, Burden of supporting the role of immunotherapy in subsets of EnCa [26,27].Cur- nonsynonymous mutations among TCGA cancers and candidate immune check- rently there is no published prospective data regarding the use of anti- point inhibitor responses, Cancer Res. 76 (13) (2016) 3767–3772. PD-1/PD-L1 in EnCa; however, trials are ongoing and recently the [8] I.C. van Gool, F.A. Eggink, L. Freeman-Mills, E. Stelloo, E. Marchi, M. de Bruyn, et al., POLE proofreading mutations elicit an antitumor immune response in endometrial phase Ib Keynote-028 (NCT02054806) trial reported a 13% response cancer, Clin. Cancer Res. 21 (14) (2015) 3347–3355. rate with acceptable toxicity in 24 women with endometrial carcinoma [9] V. Lee, A. Murphy, D.T. Le, L.A. Diaz Jr., Mismatch repair deficiency and response to [28]. In addition to solid tumor trials that include patients with EnCa, immune checkpoint blockade, Oncologist 21 (10) (2016) 1200–1211. fi fi [10] T. Zhang, J. Xie, S. Arai, L. Wang, X. Shi, N. Shi, et al., The ef cacy and safety of anti- there are several phase II trials speci cally investigating checkpoint in- PD-1/PD-L1 antibodies for treatment of advanced or refractory cancers: a meta- hibitors in EnCa. One trial (NCT02630823) administers 2 doses of analysis, Oncotarget (2016). pembrolizumab prior to surgical resection so that tissue can undergo [11] J.R. Brahmer, S.S. Tykodi, L.Q. Chow, W.J. Hwu, S.L. Topalian, P. Hwu, et al., Safety and fi activity of anti-PD-L1 antibody in patients with advanced cancer, N. Engl. J. Med. 366 immune analysis. For those patients whose pathology con rms high- (26) (2012) 2455–2465. risk features or advanced disease, pembrolizumab will be given every [12] D.M. Pardoll, The blockade of immune checkpoints in cancer immunotherapy, Nat. 3 weeks after the completion of adjuvant therapy for a maximum of 4 Rev. Cancer 12 (4) (2012) 252–264. [13] G.K. Philips, M. Atkins, Therapeutic uses of anti-PD-1 and anti-PD-L1 antibodies, Int. doses post-surgery. Another trial (NCT02549209) of patients with ad- Immunol. 27 (1) (2015) 39–46. vanced or recurrent EnCA administers pembrolizumab in conjunction [14] D.T. Le, J.N. Uram, H. Wang, B.R. Bartlett, H. Kemberling, A.D. Eyring, et al., PD-1 with carboplatin and paclitaxel. Neither of these trials utilize PD-1/PD- blockade in tumors with mismatch-repair deficiency, N. Engl. J. Med. 372 (26) – L1 status or mutational burden as a biomarker for enrollment. However, (2015) 2509 2520. [15] W. Ma, B.M. Gilligan, J. Yuan, T. Li, Current status and perspectives in translational trial NCT02912572 enrolls patients with MSS, MSI or POLE-mutated re- biomarker research for PD-1/PD-L1 immune checkpoint blockade therapy, J. current or persistent endometrial cancer for treatment with Avelumab, Hematol. Oncol. 9 (1) (2016) 47. a new anti-PD-L1 monoclonal antibody for planned biomarker analysis. [16] M. Fan, Q. Zhuang, Y. Chen, T. Ding, H. Yao, L. Chen, et al., B7-H4 expression is correlated with tumor progression and clinical outcome in urothelial cell carcinoma, Our data demonstrating that EnCa tumors with high grade histology Int. J. Clin. Exp. Pathol. 7 (10) (2014) 6768–6775. and MSI exhibit meaningful PD-L1 protein expression, alongside corre- [17] K.A. Schalper, D. Carvajal-Hausdorf, J. McLaughlin, M. Altan, V. Velcheti, P. Gaule, sponding TCGA RNA profiling data correlating PD-L1 gene expression et al., Differential expression and significance of PD-L1, IDO-1, and B7-H4 in fi human lung cancer, Clin. Cancer Res. (2016). with T cell in ltration, further support the use of immune checkpoint in- [18] J.B. Smith, C. Stashwick, D.J. Powell Jr., B7-H4 as a potential target for immunother- hibition in the treatment of subsets of women with EnCa. Our observa- apy for gynecologic cancers: a closer look, Gynecol. Oncol. 134 (1) (2014) 181–189. tion that there was a positive association between the expression of B7- [19] L. Liang, Y. Jiang, J.S. Chen, N. Niu, J. Piao, J. Ning, et al., B7-H4 expression in ovarian serous carcinoma: a study of 306 cases, Hum. Pathol. 57 (2016) 1–6. H4 and PD-L1, most notably in high grade tumors, raises the possibility [20] H. Xu, X. Chen, M. Tao, K. Chen, C. Chen, G. Xu, et al., B7-H3 and B7-H4 are indepen- that combination immunotherapy to address multiple family members dent predictors of a poor prognosis in patients with pancreatic cancer, Oncol. Lett. of checkpoint proteins simultaneously may be successful in high grade 11 (3) (2016) 1841–1846. [21] L. Liu, D. Li, S. Chen, R. Zhao, D. Pang, D. Li, et al., B7-H4 expression in human infil- tumors, which account for the majority of recurrence and death from trating ductal carcinomaassociated macrophages, Mol. Med. Rep. 14 (3) (2016) EnCa. The concept of combination immune therapies that target multi- 2135–2142. ple checkpoints has been reported on a limited basis with the use of [22] C. Chen, Q.X. Qu, Y. Shen, C.Y. Mu, Y.B. Zhu, X.G. Zhang, et al., Induced expression of Nivolumab and Ipilimunmab with significant improvement in response B7-H4 on the surface of lung cancer cell by the tumor-associated macrophages: a potential mechanism of immune escape, Cancer Lett. 317 (1) (2012) 99–105. rates in melanoma [29,30]. These data suggest that additional members [23] B.E. Howitt, S.A. Shukla, L.M. Sholl, L.L. Ritterhouse, J.C. Watkins, S. Rodig, et al., As- of the B7 family are expressed in the histologic spectrum of EnCa, and sociation of Polymerase e-mutated and microsatellite-instable endometrial cancers fi suggest that targeting additional checkpoints may improve responses with neoantigen load, number of tumor-in ltrating lymphocytes, and expression of PD-1andPD-L1,JAMAOncol.1(9)(2015)1319–1323. in non-MSI cohorts to improve the response rates. These preliminary [24] A. Vanderstraeten, C. Luyten, G. Verbist, S. Tuyaerts, F. Amant, Mapping the immu- data demonstrate an immune landscape that provides pre-clinical ratio- nosuppressive environment in uterine tumors: implications for immunotherapy, nale for the use of PD-1/PD-L1 inhibition in EnCa, particularly in those Cancer Immunol. Immunother. 63 (6) (2014) 545–557. [25] T. Miyatake, B. Tringler, W. Liu, S.H. Liu, J. Papkoff, T. Enomoto, et al., B7-H4 (DD- tumors with MSI and high grade. O110) is overexpressed in high risk uterine endometrioid adenocarcinomas and inversely correlated with tumor T-cell infiltration, Gynecol. Oncol. 106 (1) (2007) 119–127. Financial support [26] A.D. Santin, S. Bellone, N. Buza, J. Choi, P.E. Schwartz, J. Schlessinger, et al., Regression of chemotherapy-resistant polymerase epsilon (POLE) ultra-mutated and MSH6 hyper-mutated endometrial tumors with nivolumab, Clin. Cancer Res. 22 (23) This research was funded by the Advanced Medical Research Foun- (2016) 5682–5687. dation (BRR), Vincent Memorial Research Funds (BRR), Jounce Thera- [27] J.M. Mehnert, A. Panda, H. Zhong, K. Hirshfield, S. Damare, K. Lane, et al., Immune ac- peutics, Inc. tivation and response to pembrolizumab in POLE-mutant endometrial cancer, J. Clin. Invest. 126 (6) (2016) 2334–2340. [28] P.A. Ott, Y. Bang, D. Berton-Rigaud, E. Elez, M.J. Pishvaian, H.S. Rugo, et al., Pembrolizumab in advanced endometrial cancer: preliminary results from the References phase Ib KEYNOTE-028 study, J. Clin. Oncol. 34 (2016) (Suppl.; abstr 5581). [29] F.S. Hodi, J. Chesney, A.C. Pavlick, C. Robert, K.F. Grossmann, D.F. McDermott, et al., [1] R.L. Siegel, K.D. Miller, A. Jemal, Cancer statistics, 2016, CA Cancer J. Clin. 66 (1) Combined nivolumab and ipilimumab versus ipilimumab alone in patients with ad- (2016) 7–30. vanced melanoma: 2-year overall survival outcomes in a multicentre, randomised, [2] J.V. Bokhman, Two pathogenetic types of endometrial carcinoma, Gynecol. Oncol. 15 controlled, phase 2 trial, Lancet Oncol. 17 (11) (2016) 1558–1568. (1) (1983) 10–17. [30] M.S. Carlino, G.V. Long, Ipilimumab combined with nivolumab: a standard of care [3] D.M. Boruta II, P.A. Gehrig, A.N. Fader, A.B. Olawaiye, Management of women with for the treatment of advanced melanoma? Clin. Cancer Res. 22 (16) (2016) uterine papillary serous cancer: a Society of Gynecologic Oncology (SGO) review, 3992–3998. Gynecol. Oncol. 115 (1) (2009) 142–153.