Association of PD-L1 and PD-L2 expression and tumor-infiltrating lymphocytes in BRAF V600E-mutated metastatic colorectal cancer: GI-SCREEN post-hoc analysis

Abstract

Background

The programmed cell death protein 1 (PD-1)/programmed cell death-ligand 1 (PD-L1)/programmed cell death-ligand 2 (PD-L2) axis is responsible for cancer immune escape, which facilitates disease progression. However, the role of PD-L1 and PD-L2 and tumor-infiltrating lymphocytes (TILs) in metastatic colorectal cancer (mCRC) has not been studied.

Materials and methods

We conducted a post-hoc analysis of the Nationwide Cancer Genome Screening Project GI-SCREEN in mCRC. PD-L1 (22C3) and PD-L2 (MEB123.3G2.038) expression in formalin-fixed paraffin-embedded tumor samples was centrally assessed by immunohistochemical assays. TILs were morphologically evaluated using hematoxylin and eosin staining. Clinical information was extracted from the GI-SCREEN database. Inclusion of patients with BRAF V600E mutation was prioritized.

Results

Two hundred patients with mCRC (median age 65 years and 116 males) were included in the study. Genomic testing identified RAS mutations in 87 (44%) patients, BRAF V600E mutations in 27 (14%), and microsatellite instability-high status in 8 (4%). Positivity of PD-L1 and PD-L2 was 11% and 47% on tumor cells (TC) and 0% and 64% on immune cells, respectively, and that was associated with the presence of TILs (P = 0.011 for PD-L1, 0.024 for PD-L2). PD-L1+ TC was significantly more frequent in BRAF V600E-mutated tumors (P = 0.03). Even in microsatellite stable tumors, BRAF V600E-mutated tumors were significantly associated with higher expression of PD-L1 on TC than BRAF wild-type (25% versus 8%, P = 0.02).

Conclusions

Our study showed a distinct pattern of PD-L1 expression on TC of patients with BRAF V600E-mutated mCRC, which could be a potential therapeutic target for PD-1 blockade.

Highlights

• •PD-L1/PD-L2 expressions in tumor and immune cells were associated with TILs in mCRC.
• •PD-L1 expression was associated with better prognosis in mCRC.
• •BRAF V600E-mutated mCRC was likely to express PD-L1, regardless of MSI status.

Introduction

Colorectal cancer (CRC) is the third most diagnosed and the second leading cause of cancer-related mortality worldwide, and its occurrence is still on the rise.¹ Prognosis of unresectable metastatic CRC (mCRC) has improved with the introduction of new drugs, including molecular-targeted therapies; however, the cure is still difficult with systemic therapy alone.

Immune checkpoint inhibitors (ICIs) targeting programmed cell death protein 1 (PD-1)/programmed cell death-ligand 1 (PD-L1) have dramatically changed the therapeutic paradigms across multiple cancer subtypes, including melanoma, non-small-cell lung cancer (NSCLC), and renal cell carcinomas. In patients with mCRC, the efficacy of ICIs is limited to those with microsatellite instability-high (MSI-H) or tumor mutation burden-high (TMB-H);2, 3, 4 however, there is no suitable marker yet that satisfactorily predicts the response to ICIs in microsatellite stable (MSS) CRC. Therefore, predictive biomarkers for PD-1 blockade are urgently required.

Engagement of PD-1 with either of its two membrane-bound ligands, PD-L1 or PD-L2, suppresses immune responses and promotes self-tolerance.5, 6, 7 PD-L1 expression and tumor-infiltrating lymphocyte (TIL) counts serve as biomarkers of ICI response, and a strong correlation exists between these two; however, the functional role of PD-L2 in cancer cells has been scarcely investigated.^8,9 Although the structure of PD-L2 is similar to that of PD-L1, the binding affinity between PD-L2 and PD-1 is two-fold to sixfold higher than that between PD-L1 and PD-1,¹⁰ which indicates that PD-L2 is an important molecule in immune escape, since the strong interaction inhibits cytokine secretion and T-cell proliferation.¹¹

Preclinical studies have shown that increased PD-L1 expression promoted by phosphorylation of extracellular signal-regulated kinase in BRAF V600E-mutated CRCs can accelerate cell cycle processes and promote CRC cell proliferation; BRAF V600E-mutated CRC cells with PD-L1 nuclear translocation signaling were found to exhibit stronger proliferative activity.^12,13 Recently, clinical efficacy of BRAF inhibitor plus anti-epidermal growth factor receptor (EGFR) monoclonal antibody (mAb) plus anti-PD-1 mAb combination therapy in BRAF V600E-mutated MSS mCRC was reported in an early clinical trial.¹⁴

The current study aimed to conduct an analysis of PD-L1/PD-L2 protein expression as well as TILs in patients with mCRC, using survival data, to elucidate the association of PD-L1/PD-L2 expression and TILs with clinical outcomes. In addition, we examined the association across PD-L1/PD-L2 expression, genomic abnormalities, such as RAS/BRAF V600E/PIK3CA mutations, and MSI status.

Materials and methods

Study population

This was a post-hoc analysis using residual formalin-fixed paraffin-embedded tumor samples, as a sub-study of the GI-SCREEN 2013-01-CRC, profiling cancer-related genomic abnormalities. The sub-study included patients with mCRC, enrolled between February 2014 and January 2015, who met the following criteria: (i) consented to the secondary use of their data and specimens at the time of consent for GI-SCREEN 2013-01-CRC and have not withdrawn their consent; (ii) new and sufficient specimens, such as either unstained thin slides or excess cancer cell-derived DNA, to examine PD-L1, PD-L2, and MSI status; and (iii) patients with BRAF V600E were prioritized for inclusion because of funding limitation for immunohistochemical staining of PD-L1 and PD-L2 protein expression. Clinical information, including biomarker status and overall survival (OS), was extracted from the GI-SCREEN database. The current study was conducted in accordance with the Declaration of Helsinki and the Japanese Ethical Guidelines for Medical and Health Research Involving Human Subjects. All study protocols were approved by the institutional review board of each participating institution and registered in the University Hospital Medical Information Network (UMIN) Clinical Trials Registry (UMIN000016343).

PD-L1 and PD-L2 protein expression

Consecutive 4-μm-thick tissue sections were prepared for hematoxylin and eosin (HE) staining and immunohistochemical staining of PD-L1 and PD-L2. Immunostaining was carried out using an anti-PD-L1 antibody (clone 22C3, Merck, Palo Alto, CA)¹⁵ and mouse anti-human PD-L2 proprietary antibody (clone MEB123.3G2.038)¹⁶ as primary antibodies. PD-L1 and PD-L2 expression was assessed by the pathologists at Qualtek Laboratories and NeoGenomics Laboratories, respectively, in a blinded manner, for baseline characteristics and clinical outcome. Briefly, expression of PD-L1 and PD-L2 was analyzed separately in tumor cells (TC) and on immune cells (IC), since PD-L1 was expressed on TC and IC, while PD-L2 was expressed only on IC. With PD-L1 clone 22C3, nucleated cells with partially or fully stained plasma membrane were counted, as described previously.¹⁷ PD-L1 was evaluated as positive if either TC or IC was stained more than 1%; PD-L2 was evaluated as positive if either TC or IC was stained (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmogo.2023.08.007). For survival outcome, the correlation between each positive or negative result was analyzed.

Tumor-infiltrating lymphocytes

The abundance of TILs was assessed and scored by examining HE-stained sections (three to five 20× fields): score 0 corresponded to <1 TIL, score 1 corresponded to 1-10 TILs per field on an average, score 2 corresponded to 11-20, and score 3 corresponded to >20. For survival outcome analysis, negative TIL score was defined as score 0 and positive TIL score was defined as score 1 or greater.

Statistical analyses

Patient characteristics are summarized descriptively. Mann–Whitney U test for continuous quantitative variables and chi-square test for categorical variables were carried out. We investigated the association of PD-L1 and PD-L2 expression and TILs with OS. OS was estimated from the date of initiation of the first-line treatment to the date of death or of the last contact when patients were alive. Survival curve was prepared using the Kaplan–Meier method. For comparison between groups for OS, the log-rank test was carried out, and hazard ratio (HR) and confidence interval (CI) were estimated by the Cox proportional hazards model. All P values were two-sided, and P ≤ 0.05 was considered statistically significant. Statistical analyses were carried out with SAS Release 9.4 (SAS Institute, Inc., Cary, NC).

Results

Clinicopathological characteristics

Out of the 853 patients enrolled in the main study GI-SCREEN 2013-01-CRC, 733 agreed to the secondary use of data and had sufficient specimens available for PD-L1, PD-L2, and MSI analysis. Of those, 533 patients with BRAF wild-type were excluded because of funding limitation for immunohistochemical staining of PD-L1 and PD-L2 protein expression, and the remaining 200 patients with RAS/BRAF V600E mutation were included in this sub-study. Priority was given to enrolling BRAF V600E mutations requiring biomarker development and selected from patients with new specimens. PD-L1 staining was available in 194 patients, while PD-L2 staining was available in 182 patients (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmogo.2023.08.007).

The sub-study cohort had a median age of 65 years (range, 29-88 years), and 116 patients (58%) were male. Genomic testing identified RAS mutations in 87 (44%) patients and BRAF V600E mutations in 27 (14%) patients. MSI-H was detected in 8 (4%) out of 197 patients with MSI status (Table 1). Baseline characteristics in this sub-study population, including OS (median, 26.7 versus 26.3 months, HR = 1.027, 95% CI 0.73-1.43; P = 0.89), were generally similar to those in the main study population, except that BRAF V600E mutation rate and right-sided primary tumors were more prevalent in this sub-study than in the main study, since patients with BRAF V600E mutations were prioritized.

Table 1.

Patient characteristics

Number of patients, n (%)
Patient characteristics	This sub-study population	Main study population (GI-SCREEN 2013-01-CRC study)	P
	(n = 200)	(n = 853)
Age (years, range)	65 (29-88)	65 (17-88)	0.86a
Gender
Male	116 (58%)	503 (59%)	0.80b
Female	84 (42%)	350 (41%)
Histologyc
tub1/tub2	186 (93%)	768 (90%)
por1/por2	6 (3%)	49 (5.8%)	0.39b
Muc	7 (3.5%)	22 (2.6%)
Pap	1 (0.5%)	3 (0.4%)
Primary tumor location
Right-sided colon	81 (40.5%)	179 (25.1%)
Left-sided colorectum	119 (59.5%)	533 (74.9%)	<0.01b,c
Unknown	0	141
Gene mutation status
RAS (mutant/wild-type)	87(43.5%)/113	161(45.4%)/194	0.67b
BRAF (V600E mutant/wild-type)	27(13.5%)/173	38(4.4%)/827	<0.01b,c
PIK3CA (mutant/wild-type)	25(12.9%)/172	73(8.6%)/771	0.08b
MSI (MSI-H/MSS, MSI-L)	8 (4.0%)/189	15(1.9%)/790	0.06b
Overall survival, median months	26.7	26.3	0.89d

CRC, colorectal cancer; MSI-H, microsatellite instability-high; MSI-L, microsatellite instability-low; MSS, microsatellite stable; muc, mucinous adenocarcinoma; pap, papillary adenocarcinoma; por, poorly differentiated adenocarcinoma; tub, tubular adenocarcinoma.

^aMann–Whitney U test.

^bChi-square test.

^cJapanese Classification of Colorectal Carcinoma—Second English Edition.

^dCox proportional hazard model.

Correlation of PD-L1 and PD-L2 expression and TILs with clinicopathological features

Overall, 101 (52%) tumors were PD-L1+ and 117 (64%) were PD-L2+ among the PD-L1/PD-L2 evaluable cases (n = 194 and 182, respectively). PD-L1 was expressed on TC in 21 patients (11%) and on IC in 91 patients (47%), whereas PD-L2 was not expressed on TC at all and was expressed in IC in 117 (64%) patients. TIL abundance score was distributed as follows: score 3, 13%; 2, 21%; 1, 36%; and 0, 30%. Positivity of PD-L1 and PD-L2 was associated with the presence of TILs (P = 0.011 for PD-L1, P = 0.024 for PD-L2, Table 2). PD-L1 on TC was frequently seen in BRAF V600E-mutated tumors than in the wild-type (22% versus 9%, P = 0.03). However, PD-L1 and PD-L2 expression on TC and IC, as well as TILs, did not correlate with other baseline characteristics, including age, gender, primary tumor location, RAS, PIK3CA, and MSI status (Table 3). Of the BRAF V600E-mutated tumors, 8% (2/26) were MSI-H and the rest were MSS; even among MSS tumors, BRAF V600E-mutated ones were significantly associated with higher expression of PD-L1 on TC than BRAF wild-type tumors (25% versus 8%, P = 0.02, Table 4).

Table 2.

Correlation of TILs with PD-L1 expression on TC or IC and with PD-L2 expression on IC

	PD-L1 on TC or IC (n = 194)			PD-L2 on IC (n = 182)
	+	−	Pa	+	−	Pa
TILs
3	17	9		19	5
2	28	12	0.011	29	8	0.024
1	31	38		34	30
0	23	33		31	21
NA	2	1		4	1

Expression of PD-L1 on TC or IC and of PD-L2 on IC was associated with TILs (P = 0.011, P = 0.024, respectively).

IC, immune cells; NA, not applicable; PD-L1, programmed cell death-ligand 1; PD-L2, programmed cell death-ligand 2; TC, tumor cells; TIL, tumor-infiltrating lymphocyte.

^aChi-square test.

Table 3.

Correlation of PD-L1 and PD-L2 expression and TILs with clinicopathological features

	PD-L1+ on TC (n = 21)		PD-L1+ on IC (n = 91)		PD-L2+ on IC (n = 117)		TILs+ (n = 135)
	n	Pa	n	Pa	n	Pa	n	Pa
Age (years)
<65 (n = 93)	13	0.13	42	0.95	54	1	59	0.26
≥65 (n = 107)	8		49		63		76
Gender
Male (n = 116)	11	0.53	52	0.67	62	0.09	82	0.29
Female (n = 84)	10		39		55		53
Primary site
Right-sided colon	7	0.47	38	0.78	52	0.16	56	0.58
Left-sided colorectum	14		53		65		79
RAS status
Mutant (n = 87)	7	0.38	42	0.30	53	0.62	57	0.81
Wild-type (n = 113)	14		49		64		78
BRAF status
V600E mutant (n = 27)	6 (22%)	0.03	12	0.93	16	0.75	18	0.72
Wild-type (n = 173)	15 (9%)		79		101		117
PIK3CA status
Mutant (n = 25)	1	0.28	13	0.31	18	0.14	16	0.73
Wild-type (n = 172)	20		76		98		117
MSI status
MSI-H (n = 8)	1	0.85	4	0.84	5	0.92	5	0.92
MSS/MSI-L (n = 189)	19		85		110		128

As for PD-L2, no expression was found on TC. The PD-L1 on TC was highly seen in BRAF V600E-mutated tumors (P = 0.03), but not related to other factors including RAS and MSI status.

Bold value indicates the significant with a P-value of <0.05.

IC, immune cells; MSI-H, microsatellite instability-high; MSI-L, microsatellite instability-low; MSS, microsatellite stable; PD-L1, programmed cell death-ligand 1; PD-L2, programmed cell death-ligand 2; TC, tumor cells; TIL, tumor-infiltrating lymphocyte.

^aP, chi-square test.

Table 4.

Relationship across BRAF alteration, MSI status, and PD-L1

Number of patients, n (%)
BRAF	n	MSI	n	PD-L1 on TC	n	PD-L1 on IC	n	PD-L1	n
BRAF V600E	26	MSI-H	2	Positive	0	Positive	2	Positive	2
				Negative	2	Negative	0	Negative	0
		MSS	24	Positive	6a(25%)	Positive	10	Positive	12
				Negative	18 (75%)	Negative	14	Negative	12
BRAF WT	165	MSI-H	6	Positive	1	Positive	2	Positive	3
				Negative	5	Negative	4	Negative	3
		MSS	159	Positive	13a(8.2%)	Positive	75	Positive	82
				Negative	146 (91.8%)	Negative	84	Negative	77

Bold values indicate the significant with a P-value of <0.05.

IC, immune cells; MSI-H, microsatellite instability-high; MSS, microsatellite stable; PD-L1, programmed cell death-ligand 1; TC, tumor cells; WT, wild-type.

^aP < 0.05, chi-square test.

Survival outcome according to PD-L1/PD-L2 expression and TIL status

Patients with PD-L1+ tumors (having PD-L1 on either TC or IC) had a significantly longer OS than those with PD-L1− (median, 31.9 versus 23.5 months; HR = 0.67, 95% CI 0.46-0.99; P = 0.040) (Figure 1A), although patients with PD-L2+ tumors had OS similar to those with PD-L2− (median, 25.7 versus 34.8 months; HR = 1.13, 95% CI 0.74-1.72; P = 0.57) (Figure 1B). Patients with TIL+ tumors tended to have a longer OS than those with TIL− (median, 29.9 versus 24.6 months; HR = 0.71, 95% CI 0.48-1.06; P = 0.09) (Figure 1C). PD-L1 and PD-L2 expression was significantly correlated (P = 0.01) (Supplementary Table S1, available at https://doi.org/10.1016/j.esmogo.2023.08.007), and OS was significantly shorter in the PD-L1− and PD-L2+ tumors than in others (Supplementary Figure S3, available at https://doi.org/10.1016/j.esmogo.2023.08.007).

Figure 1.

OS basedon PD-L1, PD-L2, and TIL status. (A) Kaplan–Meier plots of OS with and without PD-L1 expression. (B) Kaplan–Meier plots of OS with and without PD-L2 expression. (C) Kaplan–Meier plots of OS with and without TIL. CI, confidence interval; HR, hazard ratio; OS, overall survival; PD-L1, programmed cell death-ligand 1; PD-L2, programmed cell death-ligand 2; TIL, tumor-infiltrating lymphocyte.

In addition, patients with MSS BRAF V600E-mutated mCRC with PD-L1+ tumors had a longer OS than those with PD-L1−, with a non-significant trend (median, 17.2 versus 8.5 months; HR = 0.51, 95% CI 0.20-1.28; P = 0.15) (Figure 2A). Among this subgroup, analysis of OS according to PD-L1 expression on TC or IC showed a trend toward longer OS for PD-L1+ than for PD-L1− in both the groups. The trend was clearer for PD-L1+ on IC, although the difference was not significant (median, 24.2 versus 11.4 months; HR = 0.66, 95% CI 0.22-1.99; P = 0.45; median, 24.2 versus 8.5 months; HR = 0.40, 95% CI 0.15-1.07; P = 0.06) (Figure 2B and C).

Figure 2.

OS by PD-L1 localization in patients with MSS and BRAF V600E-mutated mCRC. (A) Kaplan–Meier plot of OS with and without overall PD-L1 expression. (B) Kaplan–Meier plot of OS with and without PD-L1 expression on TC. (C) Kaplan–Meier plot of OS with and without PD-L1 expression on IC. CI, confidence interval; HR, hazard ratio; IC, immune cells; mCRC; metastatic colorectal cancer; MSS, microsatellite stable; OS, overall survival; PD-L1, programmed cell death-ligand 1; TC, tumor cells.

Discussion

Our study evaluated PD-L1 and PD-L2 expression and TILs, according to the key biomarker status, namely BRAF V600E, in patients with mCRC. The study showed that PD-L1 expression was associated with better prognosis in mCRC while PD-L2 expression was not; OS was significantly shorter in the PD-L1− and PD-L2+ tumors than in others. Furthermore, patients with BRAF V600E-mutated mCRC were likely to express PD-L1 regardless of MSI status, which was related to a longer OS. The results suggested the biological relevance of PD-L1 expression with the prognosis of patients with mCRC, potentially supporting the ICI strategy in mCRC, especially in the BRAF V600E-mutated subtype.

In this study, PD-L1 was expressed on TC in 11% and on IC in 47% of patients with mCRC, consistent with previous studies that reported a range from 5.5% to 65% on TC and from 26% to 76% on IC.18, 19, 20, 21, 22 In contrast, PD-L2 was expressed on IC in 64% of patients, although PD-L2 expression was reported to be 38.7% on IC in CRC by Wang et al.²³ This inconsistent result might depend on the antibody used to recognize PD-L2, and on how staining intensity was assessed.⁹ However, since PD-L2 expression in CRC has scarcely been investigated, further studies would be required to clarify the potential clinical value of PD-L2 expression.

Next, we investigated the correlation of PD-L1 and PD-L2 expression with baseline patient characteristics and key genomic abnormalities, such as RAS, BRAF V600E PIK3CA, and MSI status; PD-L1 expression on TC was significantly correlated with BRAF V600E-mutated tumors than with wild-type (22% versus 9.0%, P = 0.03), but not with baseline patient characteristics and other genomic abnormalities. Even among MSS tumors, BRAF V600E-mutated ones were significantly associated with higher expression of PD-L1 on TC than wild-type tumors (25% versus 8.2%; P = 0.02). Although positive correlation of PD-L1 expression with BRAF V600E-mutated mCRC has been reported in previous studies,²⁴ the impact of MSI-H, which often overlaps with BRAF V600E mutation in mCRC, in the correlation was not clear. Our study revealed that PD-L1 positivity was higher in BRAF V600E-mutated mCRC, regardless of MSI status. Preclinical studies have suggested the regulation of PD-L1 expression via mitogen-activated protein kinaseーextracellular signal-regulated kinase signaling and up-regulation of the immunosuppressive ligand PD-L1 through interferon gamma- and T helper type 1 cell-associated expression activation as a mechanism of PD-L1 expression in BRAF V600E-mutated mCRC.^12,25 Furthermore, patients with MSS BRAF V600E-mutated mCRC with PD-L1+ tumors were likely to have a longer OS regardless of PD-L1 positivity on TC or IC. These findings might support the underlying mechanism of action, at least in part, for the promising clinical efficacy of BRAF inhibitor plus anti-EGFR mAb plus anti-PD-1 mAb combination therapy in BRAF V600E-mutated mCRC with MSS observed in an early clinical trial.¹⁴

PD-L1 expression was significantly associated with favorable clinical outcomes in mCRC, and TIL-positivity showed the same trend. In contrast, although PD-L2 expression was not associated with poor prognosis, PD-L2 expression without PD-L1 was associated with poor prognosis. Previous observational studies and meta-analyses have shown that PD-L1 or PD-L2 expression is associated with poor prognosis in CRCs, predominantly in early-stage resectable disease.^26,27 Contrary to the previous studies, our current study suggested a positive impact of PD-L1 expression on the prognosis of patients with mCRC. The discordance of the prognostic impact of PD-L1 expression between early and metastatic disease was in line with that in NSCLC. A meta-analysis of 11 444 patients with NSCLC had shown that increased PD-L1 expression was associated with poor prognosis;²⁸ another meta-analysis of 4378 patients treated with chemotherapy for metastatic NSCLC revealed high PD-L1-expressing tumors to be associated with longer OS and improved objective response rate when treated with chemotherapy.²⁹ PD-L1 has been reported as a response biomarker for cytotoxic chemotherapy using pemetrexed, indicating that PD-L1 could promote response to chemotherapy; however, it remains to be fully validated.³⁰

Overlap of the presence of TILs may contribute to longer OS in patients with PD-L1+ tumors. Indeed, positivity of PD-L1 and PD-L2 was associated with TILs (P = 0.011 for PD-L1, P = 0.024 for PD-L2), which was consistent with previous reports.³¹ TILs may reflect attempts by the immune system to counter tumor responses, potentially leading to better prognosis. However, a further assessment of the subtype of TILs, such as B cells, T cells, natural killer cells, macrophages, dendritic cells, and neutrophils, should be conducted to evaluate the relationship between TILs and prognosis.

It would be important to note the limitations of the present study. Firstly, the sample size was limited for this analysis. However, OS of patients included in this study was not different from that of the overall population who were enrolled before initiation of the first-line treatment. This suggested that the patients might represent the overall population with minimized selection bias. Nevertheless, since the number of patients with each biomarker was limited, because of the small sample size, the association between PD-L1/PD-L2 status and biomarkers should be evaluated in a future study. Secondly, this study used a prototype PD-L1 assay for scoring TC and IC, which could be one possible limitation of PD-L1 assessment. Finally, although TILs were evaluated based on morphology alone in this study, TIL subtypes should be assessed using immunohistochemistry analysis of immune biomarkers or other modalities, such as single-cell analysis, in future studies.

Conclusions

Our study showed a distinct pattern of PD-L1 expression on TCs of BRAF V600E-mutated mCRC, which would require further investigation as a potential therapeutic target for PD-1 blockade.