If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Studies have reported the relevance of immune phenotype, or presence of cluster of differentiation 8 (CD8)-positive tumour-infiltrating lymphocytes, to the anti-tumour efficacy of checkpoint inhibitors and to prognosis. The multicentre, retrospective ARCHERY study (UMIN000034131) collected tissue samples from Japanese patients with recurrent or metastatic renal cell carcinoma (RCC) who received systemic therapy between 2010 and 2015. In this exploratory analysis, the prognostic impact of immune phenotype and PD-L1 expression (separately and combined) was investigated using 770 surgical specimens and outcomes from patients enrolled in ARCHERY. A key objective was to determine overall survival (OS), defined as time from nephrectomy to death from any cause, by immune and PD-L1 subgroups.
The median OS by immune phenotype was 28.8, 57.3, and 63.4 months in patients with inflamed, excluded, and desert tumours, respectively [hazard ratio (95% CI): inflamed 1.78 (1.27–2.49); excluded 1.08 (0.89–1.30); desert as reference]. PD-L1 positivity by SP142 showed a strong association with immune phenotype; 88.1%, 61.9%, and 8.7% of PD-L1-positive patients had inflamed, excluded, and desert phenotypes, respectively. PD-L1 positivity was also associated with worse OS in each phenotype, except for the inflamed phenotype (due to limited sample size in the PD-L1-negative immune inflamed subgroup; n=7). Additionally, the difference in OS by PD-L1 status was larger in the desert versus excluded phenotype [median OS in PD-L1 positive vs negative: 27.1 vs 67.2 months (desert), and 48.2 vs 78.1 months (excluded)]. Results show that PD-L1 expression was highly associated with immune phenotype, but both covariates should be evaluated when determining prognosis.
Even though CIT monotherapy has been successfully applied across a broad range of human cancers, its benefit is still limited. To aid in the development of improved CIT options for patients with cancer, it is important to understand the tumour microenvironment.
Classifying immunological phenotypes in human solid tumours based on the spatial localisation of cluster of differentiation 8 (CD8)-positive tumour-infiltrating lymphocytes (TILs) was recently advocated and considered as a framework for selecting patients and developing treatments to overcome the limited effectiveness of checkpoint inhibitor monotherapy.
One classification involves categorising tumours into three immune phenotypes: inflamed (TILs are abundant in the tumour area), excluded (TILs accumulate in the tumour but infiltration is limited), and desert (TILs are absent within the tumour and its periphery).
Several clinical trials have reported the relevance of immune phenotype, or presence of CD8-positive TILs within the tumour, to the anti-tumour efficacy of checkpoint inhibitors.
T-cell-inflamed gene-expression profile, programmed death ligand 1 expression, and tumor mutational burden predict efficacy in patients treated with pembrolizumab across 20 cancers: KEYNOTE-028.
In IMvigor210, pretreatment samples from immune inflamed patients with metastatic urothelial cancer who were treated with atezolizumab [programed death-ligand 1 (PD-L1) inhibitor] had significantly higher CD8-positive T-effector cell scores in those who were complete or partial responders than those with stable or progressive disease.
KEYNOTE-028 reported higher response rates and longer progression-free survival in 20 different types of PD-L1-positive advanced solid tumours with a high T-cell-inflamed gene expression profile, PD-L1 expression, tumour mutational burden before treatment with pembrolizumab [programmed death-1 (PD-1) inhibitor], or combination thereof.
T-cell-inflamed gene-expression profile, programmed death ligand 1 expression, and tumor mutational burden predict efficacy in patients treated with pembrolizumab across 20 cancers: KEYNOTE-028.
In a majority of solid tumours [e.g., melanoma, triple negative breast cancer, non-small cell lung cancer (NSCLC), head and neck cancer, and colorectal cancer], the presence of TILs is associated with good prognosis. However, multiple studies have demonstrated that TIL presence is associated with poor prognosis in patients with renal cell carcinoma (RCC).
Previous RCC studies have reported that high CD8-positive T-cell infiltration, expression of several checkpoints (PD-1, PD-L1/2, and T-cell immunoglobulin and mucin-domain containing-3), and genomic instability were correlated with poor prognosis and aggressive phenotype defined by high nuclear grade [such as Fuhrman grade and World Health Organization/International Society of Urological Pathology (WHO/ISUP) grade], larger tumour size at diagnosis, and advanced clinical stage.
Hence, understanding the relationship between immune checkpoints and immune phenotypes may be critical for consideration in future RCC treatment strategies.
The ARCHERY study investigated the prognostic significance of PD-L1 in Japanese patients with recurrent or metastatic RCC who started systemic therapy in 2010–2015.
Enrolment was limited to patients who had undergone nephrectomy and had surgical specimens available for pathological and immunohistochemistry (IHC) evaluation. The median overall survival (OS) after first line therapy, defined as the time from the initiation of first line therapy to death from any cause, for PD-L1-positive patients was 30.9 months (95% CI 25.5–35.7) and 37.5 months (95% CI 34.0–42.6) for PD-L1-negative patients [HR 1.04; 90% CI 0.89–1.22; p=0.65; stratified by Memorial Sloan Kettering Cancer Center (MSKCC) risk and liver metastases]. This retrospective study showed that PD-L1 positivity was associated with poor prognostic factors, particularly MSKCC risk status, suggesting that PD-L1 status was not an independent prognostic factor in recurrent/metastatic RCC during the study period.
Since previous studies have also investigated the relationship between immune phenotype, in addition to PD-L1, and prognosis in RCC,
CD8 phenotype was also evaluated for all patients enrolled in the ARCHERY study. This exploratory analysis focused on the prognostic impact of both CD8-infiltration phenotype and PD-L1 status in RCC. As these factors were evaluated at the time of surgery, we investigated their associations with OS after nephrectomy. In addition, the association between PD-L1 expression and immune phenotype was investigated.
It is notable that in lung cancer, several assays that comprise different antibodies, scoring criteria, and clinical cut-offs have been used to assess PD-L1 expression in the clinical setting.
Each assay potentially displays unique staining properties that could prohibit the interchangeability of their clinical use. Hence, several investigations have evaluated the comparability of different PD-L1 IHC assays and their potential interchangeability in clinical practice. The Blueprint PD-L1 IHC Comparability Project investigated five trial validated assays (22C3, 28-8, SP142, SP263, and 73-10) in NSCLC samples.
They reported that the 22C3, 28-8, and SP263 assays showed highly comparable staining, and the SP142 assay, which was used in the ARCHERY study, had less sensitivity. Although these studies were performed in lung cancer, it is of clinical relevance to investigate if they are also applicable in RCC.
Thus, in this exploratory analysis, the prognostic impact of immune phenotype (both separately and combined) and PD-L1 expression was investigated in patients who were enrolled in the ARCHERY study. A concordance analysis between SP142 and SP263 in patients with metastatic RCC in ARCHERY was also conducted to determine their potential interchangeability in the clinical setting.
Patients and methods
Patients
A total of 770 surgical specimens were obtained from patients enrolled in the ARCHERY study.
Briefly, samples were collected between November 2018 and June 2019 from 29 Japanese study sites that provided guideline based, standard of care treatment. Eligible patients had distant metastasis after or at the time of nephrectomy. All enrolled patients were aged ≥20 years at the time of their nephrectomy, and they received systemic therapy for recurrent or metastatic RCC between January 2010 and December 2015. Exclusion criteria were coexisting malignancies (from nephrectomy to death) or first line treatment with anti-cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) or anti-PD-(L)1 for recurrent or metastatic RCC. Although PD-L1 has not been established as a biomarker for checkpoint inhibitors in RCC,
several Phase III studies showed longer OS and/or progression-free surivival with PD-L1 inhibitors versus sunitinib, a tyrosine kinase inhibitor, in patients who were PD-L1 positive.
Thus, to ensure that the interpretation of the results would not be complicated by PD-L1 status-dependent treatment effect, we excluded patients who received first line treatment with checkpoint inhibitors.
The biomarker evaluable population comprised patients whose surgical specimens were analysed by both VENTANA SP142 and SP263 assays. This study is registered under UMIN (UMIN000034131) and was performed after approval by each institutional review board of the 29 study sites. In addition, approval was obtained from the institutional review board of MINS, a non-profit organisation.
Study design
ARCHERY was a multicentre, retrospective study that compared OS by PD-L1 expression status in patients with recurrent or metastatic RCC who had received systemic therapy to determine its prognostic effect. In this exploratory analysis, the prognostic properties of PD-L1 expression and immune phenotype (characterised by CD8-positive TILs) were analysed in patients with both localised and metastatic RCC.
One of the objectives was OS after nephrectomy, which was defined as the time from the date of nephrectomy to the date of death from any cause, in subgroup analyses that compared data from patients with inflamed vs excluded vs desert immune phenotype (desert phenotype as reference), patients in six subgroups based on PD-L1 status and immune phenotype [PD-L1-positive with each immune phenotype (inflamed vs excluded vs desert) vs PD-L1-negative with each immune phenotype (inflamed vs excluded vs desert); PD-L1-negative with desert phenotype as reference, and PD-L1-positive vs PD-L1-negative patients (PD-L1-negative as reference)]. Other exploratory analyses included the association between immune phenotype and PD-L1 expression, concordance between SP142 and SP263 (with SP142 as reference), and subgroup analysis of OS by PD-L1 status after each nephrectomy procedure (radical nephrectomy and cytoreductive nephrectomy).
Pathology and IHC
Representative formalin-fixed, paraffin-embedded samples were selected by pathologists in each institution based on a selection guidance that included sufficient tumour cells, high nuclear grade, presence of sarcomatoid component, presence of necrosis, and high tumour-associated lymphocytic infiltration (desert < excluded < inflamed). Each of these criteria were re-evaluated and scored by a central pathologist (T. Tsuzuki). PD-L1 expression was evaluated by IHC using the VENTANA SP142 assay (Ventana Medical Systems, USA) and SP263 assay (Ventana Medical Systems, in the biomarker-evaluable population) by two pathologists (T. Tsuzuki and C. Ohe) independently. All discordant cases were re-evaluated by these two pathologists and scored after they reached consensus. Based on the PD-L1 expression levels on immune cells (IC) using SP142 and SP263, patients were classified as either PD-L1 negative (IC0; IC <1%) or PD-L1 positive [further divided into IC1 (IC ≥1% but <5%), IC2 (IC ≥5% but <10%), or IC3 (IC ≥10%)]. We evaluated three distinct immunological phenotypes based on previous studies
: inflamed (many CD8-positive T cells diffusely infiltrate into tumour cell nests), excluded (CD8-positive T cells infiltrate around tumour cell nests but not into them, and desert (no or few CD8-positive T cells infiltrate around and/or into tumour cell nests). Immunological phenotyping was conducted using slides stained with haematoxylin and eosin and confirmed using CD8 immunostaining (Supplementary Fig. 1, Appendix A).
after complementary log-log transformation, and HRs were estimated using the Cox proportional hazards model. Multivariable analysis was conducted using the Cox proportional hazards model, with OS after nephrectomy as the objective variable and PD-L1, immune phenotype, and clinical stage as explanatory variables.
Results
Patient characteristics
Of 770 surgical specimens obtained, 381 samples were obtained from radical nephrectomy and 389 from cytoreductive nephrectomy (Table 1). The prevalence of CD8-positive TILs (desert < excluded < inflamed) was associated with poorer clinicopathological characteristics. The percentages of patients in clinical stage IV at the time of initial diagnosis with inflamed, excluded, and desert phenotypes were 69.5%, 57.9%, and 47.1%, respectively.
Table 1Baseline clinicopathological characteristics by immune phenotype
None of the patients received neoadjuvant chemotherapy.
0
6 (1.6)
11 (3.3)
17 (2.2)
No
59 (100)
371 (98.1)
322 (96.7)
752 (97.7)
Unknown
0
1 (0.3)
0
1 (0.1)
Pathological characteristics
Histology
Clear cell RCC
52 (88.1)
367 (97.1)
277 (83.2)
696 (90.4)
Papillary RCC
0
4 (1.1)
31 (9.3)
35 (4.5)
Chromophobe RCC
1 (1.7)
0
6 (1.8)
7 (0.9)
Sarcomatoid carcinoma
4 (6.8)
2 (0.5)
1 (0.3)
7 (0.9)
Others
2 (3.4)
5 (1.3)
18 (5.4)
25 (3.2)
Fuhrman grade
1
0
1 (0.3)
0
1 (0.1)
2
7 (11.9)
94 (24.9)
152 (45.6)
253 (32.9)
3
21 (35.6)
214 (56.6)
139 (41.7)
374 (48.6)
4
30 (50.8)
69 (18.3)
39 (11.7)
138 (17.9)
Indeterminable
1 (1.7)
0
3 (0.9)
4 (0.5)
WHO/ISUP grade
1
0
1 (0.3)
0
1 (0.1)
2
7 (11.9)
120 (31.7)
173 (52.0)
300 (39.0)
3
17 (28.8)
159 (42.1)
108 (32.4)
284 (36.9)
4
34 (57.6)
98 (25.9)
49 (14.7)
181 (23.5)
Indeterminable
1 (1.7)
0
3 (0.9)
4 (0.5)
Necrosis
Present
34 (57.6)
195 (51.6)
102 (30.6)
331 (43.0)
Absent
25 (42.4)
182 (48.1)
230 (69.1)
437 (56.8)
Indeterminable
0
1 (0.3)
1 (0.3)
2 (0.3)
Lymphovascular invasion
Present
18 (30.5)
95 (25.1)
76 (22.8)
189 (24.5)
Absent
38 (64.4)
266 (70.4)
235 (70.6)
539 (70.0)
Indeterminable
3 (5.1)
17 (4.5)
22 (6.6)
42 (5.5)
Sarcomatoid component
Present
26 (44.1)
37 (9.8)
24 (7.2)
87 (11.3)
Absent
33 (55.9)
341 (90.2)
309 (92.8)
683 (88.7)
Indeterminable
0
0
0
0
Growth pattern
Expansive
18 (30.5)
143 (37.8)
117 (35.1)
278 (36.1)
Infiltrative
18 (30.5)
94 (24.9)
71 (21.3)
183 (23.8)
Indeterminable
23 (39.0)
141 (37.3)
145 (43.5)
309 (40.1)
IC, immune cells; PD-L1, programmed death-ligand 1; RCC, renal cell carcinoma; WHO/ISUP, World Health Organization/International Society of Urological Pathology.
a Defined as IC1/2/3.
b Defined as IC0.
c None of the patients received neoadjuvant chemotherapy.
Histology also showed association with immune phenotype. Thirty-one of 35 cases with papillary cell carcinoma and six of seven cases with chromophobe RCC were categorised as desert phenotype. In addition, pathological grade was imbalanced across immune phenotypes. The majority of patients with inflamed phenotype had pathological Grade 4 disease (WHO/ISUP grade 57.6%; Fuhrman grade 50.8%). On the other hand, most immune desert patients had pathological Grade 2 disease (WHO/ISUP grade 52.0%; Fuhrman grade 45.6%).
Baseline characteristics by PD-L1 status were previously reported,
and baseline clinicopathological characteristics by PD-L1 status and immune phenotype are shown in Supplementary Table 1 (Appendix A). Baseline characteristics at the time of first line therapy by immune phenotype are presented in Supplementary Table 2 (Appendix A).
Association between PD-L1 status and phenotype
PD-L1 positivity by SP142 showed strong association with immune phenotype. Among PD-L1-positive patients, 88.1%, 61.9%, and 8.7% had inflamed, excluded, and desert phenotypes, respectively (Table 1). Descriptive statistics of measured PD-L1 expression values also demonstrated similar association in the biomarker-evaluable population (n=389), in which PD-L1 expression was evaluated as a percentage of PD-L1-positive IC in the tumour (Supplementary Table 3, Appendix A). PD-L1 status in patients with WHO/ISUP Grade 4 disease by presence of sarcomatoid features are shown in Supplementary Table 4 (Appendix A).
Survival analysis
Median OS after nephrectomy by immune phenotype was 28.8, 57.3, and 63.4 months, in the inflamed, excluded, and desert phenotypes, respectively (inflamed vs desert: HR 1.78, 95% CI 1.27–2.49; excluded vs desert: HR 1.08, 95% CI 0.89–1.30; Fig. 1). The median OS after nephrectomy was 41.5 months in the PD-L1-positive subgroup and 67.8 months in the PD-L1-negative subgroup (HR 1.48; 95% CI 1.23–1.77; Supplementary Fig. 2, Appendix A). Similar observations were made for patients who underwent radical nephrectomy and cytoreductive nephrectomy (Supplementary Fig. 3, Appendix A).
Fig. 1OS after nephrectomy by immune phenotype. CI, confidence interval; HR, hazard ratio; NA, not applicable; OS, overall survival.
In the PD-L1-positive group, patients with inflamed phenotype (n=52) and desert phenotype (n=29) had worse OS than those with excluded phenotype (n=234; median OS 28.8, 27.1, and 48.2 months, respectively). In the PD-L1-negative group, immune desert (n=304) and excluded (n=144) patients showed similar OS (median OS 67.2 vs 78.1 months; HR 0.92; Table 2, Fig. 2).
Table 2Overall survival after nephrectomy by immune phenotype and PD-L1 status
PD-L1 positivity was associated with poorer OS in each phenotype, except for the inflamed phenotype (due to limited sample size in the PD-L1-negative immune inflamed subgroup; n=7). In addition, the difference in OS by PD-L1 status was larger in the desert phenotype than in the excluded phenotype [median OS in PD-L1 positive vs negative: 27.1 vs 67.2 months (desert) and 48.2 vs 78.1 months (excluded) (Table 2, Fig. 2)]. These findings were supported by the results of a multivariable analysis with OS as the objective variable (Supplementary Table 5, Appendix A).
Concordance analysis (SP142 vs SP263)
There were 389 patients in the biomarker-evaluable population. The numbers of PD-L1-positive and-negative patients, determined by both SP142 and SP263 assays, in the biomarker evaluable population are shown in Table 3. Using the SP142 assay as reference, the positive, negative, and overall percentage agreements of the two assays were 97%, 73%, and 83%, respectively.
Table 3Comparison of PD-L1-positive vs -negative expression based on SP142 and SP263 evaluation
In the biomarker evaluable population, the median OS after nephrectomy in the PD-L1-positive and -negative subgroups by SP142 was 54.9 and 78.1 months, respectively (HR 1.41). The median OS after nephrectomy in the PD-L1-positive and -negative subgroups by SP263 was 54.9 and 85.6 months, respectively (HR 1.58; Supplementary Fig. 4, Appendix A).
Association between PD-L1 expression and phenotype by PD-L1 status (SP142 and SP263)
The association between PD-L1 expression (percentage of PD-L1 on PD-L1-positive IC by both SP142 and SP263) and phenotype was observed in the biomarker evaluable population. The descriptive statistics of measured PD-L1 levels on IC by SP142 and SP263 showed similar trends (Supplementary Table 3, Appendix A). Median PD-L1 expression on IC by SP142 was 5.0%, 1.0%, and 0% in immune inflamed, excluded, and desert tumours, respectively. Median PD-L1 level on IC by SP263 was 7.0%, 2.0%, and 0% for inflamed, excluded, and desert phenotypes, respectively. The results based on descriptive statistics were representative of the distribution of PD-L1 values on IC shown in the box plots (Supplementary Fig. 5, Appendix A). In this study, the association of PD-L1 expression on tumour cells (TC) with phenotype could not be evaluated by both SP142 and SP263 because most of the PD-L1 levels on TC were 0% [SP142: 366/389 (94.1%); SP263: 312/389 (80.2%)] (Supplementary Table 3 and Supplementary Fig. 5B,D, Appendix A).
Discussion
Immune phenotype, PD-L1 expression, and baseline characteristics
Previously, the ARCHERY study reported a higher prevalence of PD-L1 positivity in patients with poor prognostic features (i.e., ‘poor’ MSKCC or International Metastatic RCC Database Consortium risk) and high-risk pathological features (higher clinical stage, higher nuclear grade, and sarcomatoid features).
In this exploratory analysis, PD-L1 expression was highly associated with immune phenotype (Table 1). In addition, PD-L1 expression in IC by both SP142 and SP263 was associated with phenotype in the biomarker-evaluable population (Supplementary Table 3 and Supplementary Fig. 5, Appendix A). Immune inflamed patients had the most aggressive prognostic features (higher nuclear grade and presence of sarcomatoid components) and showed highest PD-L1 positivity among the three phenotypes. Thus, the results from this exploratory analysis are consistent with those of earlier studies that reported higher PD-L1 expression and higher density of CD8-positive cell infiltrate in RCC with higher nuclear grade or presence of sarcomatoid components.
Efficacy and correlative analyses of avelumab plus axitinib versus sunitinib in sarcomatoid renal cell carcinoma: post hoc analysis of a randomized clinical trial.
Atezolizumab plus bevacizumab versus sunitinib for patients with untreated metastatic renal cell carcinoma and sarcomatoid features: a prespecified subgroup analysis of the IMmotion151 clinical trial.
Hence, the associations among sarcomatoid components, immune phenotype, and PD-L1 expression in this analysis are consistent with those in previous studies.
Immune phenotype and histology
In this study, there was an association between immune phenotype and histology (Table 1). The majority of patients had clear cell histology (90.4%), which is consistent with the analysis of a population-based cancer registry in Japan (88.2%).
Patients with clear cell histology were classified according to the three immune phenotypes, with most having the excluded phenotype (53%; 367/696). On the other hand, most patients with papillary (89%; 31/35) and chromophobe histology (86%; 6/7) were classified as immune desert. These results are similar to those reported by a study that examined the relative fractions of 22 tumour-infiltrating immune cell types in 891 primary kidney tumour tissues using data from The Cancer Genome Atlas.
Immune phenotype, pathological grade, and prognosis
Of the three immune phenotypes, immune inflamed was associated with the worst OS and immune desert was associated with the best OS. This finding is consistent with previous studies that reported the correlation between CD8-positive T-cell infiltration and poor prognosis in patients with RCC.
Since the excluded phenotype is characterised by the presence of CD8-positive immune cells near the tumour area, immune excluded patients were expected to show an OS trend similar to that in immune inflamed patients. However, contrary to expectations, the OS rate in patients with the excluded phenotype was very similar to that in patients with the desert phenotype.
Higher pathological grade is associated with poorer prognosis in RCC and therefore has been incorporated into prognostic models for localised disease.
An outcome prediction model for patients with clear cell renal cell carcinoma treated with radical nephrectomy based on tumor stage, size, grade and necrosis: the SSIGN score.
Hence, the distribution of pathological grade in each phenotype was examined. The OS rate in patients with inflamed phenotype was worse than those in patients with excluded or desert phenotypes (Fig. 2), possibly due to an imbalance of pathological grade distribution. However, this hypothesis could not explain why immune excluded and desert patients had similar OS, because there were more patients with poorer pathological grade in the excluded phenotype subgroup than in the desert phenotype subgroup (Table 1).
Recently, the association between immunospatial profiles of intratumoural and peritumoural tissue and clinicopathological characteristics was studied in patients with clear cell RCC.
Intratumoural T cells, which are present in high numbers in immune inflamed patients, expressed higher levels of immune activation and exhaustion markers, such as PD-1 and LAG3, than T cells in the peripheral region. In addition, immune cell interaction was most frequent in the intratumoural region and was correlated with higher CD45RO expression, which predicted poor prognosis.
Immune phenotype, PD-L1 status, and prognosis
The impact of both immune phenotype and PD-L1 expression on OS was investigated. Overall, PD-L1-positive patients had shorter OS than PD-L1-negative patients after nephrectomy (Supplementary Fig. 2, Appendix A). In patients with desert and excluded phenotypes, OS was shorter in PD-L1-positive patients than in PD-L1-negative patients (Fig. 2).
Since there were more immune excluded and desert patients with poorer nuclear grade in the PD-L1-positive subgroup than the PD-L1-negative subgroup, the difference in OS by PD-L1 expression was likely due to the imbalanced distribution of nuclear grades (Supplementary Table 1, Appendix A). However, this imbalance of nuclear grade distribution cannot account for the larger difference in OS between PD-L1-positive and-negative patients observed in the desert and excluded phenotypes (Fig. 2).
Since PD-L1 is a known biomarker of resistance to tumour immunity, the shorter OS observed in the PD-L1-positive immune excluded group compared with the PD-L1-negative immune excluded group was possibly caused by immune resistance.
Despite the limited number of patients with desert phenotype, OS was the shortest in PD-L1-positive, immune desert patients, consistent with other findings. In addition, this type of tumour is likely to be resistant to PD-L1 immunotherapy.
It has previously been shown that grouping tumour-infiltrating immune cells by immune phenotype [based on expression of CD8, PD-(L)1, T-cell immunoglobulin and mucin-domain containing-3, and other biomarkers] can lead to differences in clinical outcomes.
In the PD-L1-positive group, the difference in OS between immune inflamed and excluded patients (28.8 vs 48.2 months) further emphasises the importance of CD8-positive T-cell localisation in characterising tumour immunity status.
A recent study reported that even though CD8-positive infiltration was not associated with clinical response to PD-1 blockade, highly CD8-positive T-cell-infiltrated clear-cell RCC tumours had numerous chromosomal alterations associated with response or resistance to PD-1 blockade, demonstrating the relationship between immune phenotype and PD-L1 expression that was observed.
Sixty-one patients (15.7%) were classified as PD-L1-negative by SP142 but PD-L1-positive by SP263. However, only five patients (1.3%) were categorised as PD-L1-positive by SP142 but PD-L1-negative by SP263. The overall percentage agreement was also similar to that reported in Phase 1 of the Blueprint Project. The overall percentage agreement between SP142 and SP263 using the TC1/IC1 scoring algorithm was 86.8%, whereas this study reported 83% between SP142 and SP263 using the IC1 scoring algorithm.
In the survival analysis, OS after nephrectomy was shorter in the PD-L1-positive subgroup than in the PD-L1-negative subgroup, regardless of the assay used to determine PD-L1 status (Supplementary Fig. 4, Appendix A).
Limitations
Due to the retrospective nature of the ARCHERY study, there may have been confounding factors affecting prognosis that were not measured. This study enrolled patients with recurrent or metastatic RCC who were treated with systemic therapy, excluding patients who were cured by surgery. According to statistics from the Japanese National Cancer Center, 55.5% of patients with kidney and other genitourinary cancers (except for bladder cancer) were diagnosed with localised disease and had a 5-year survival rate of 94.3%.
Center for Cancer Control and Information Services National Cancer Center Cancer survival rate data by regional cancer registration (5-year survival rate of 1993-2011 diagnosis cases).
Thus, patients from this study had more advanced disease than the real world patient population of Japan.
As this study required surgical samples, patients with metastatic RCC who were ineligible for surgery were also excluded. A multicentre, retrospective medical chart review study in Japan reported that 81.9% of patients with metastatic RCC undergo surgery; therefore, only a small percentage of the real world Japanese metastatic RCC population was excluded from this study.
In addition, the definition of OS, defined as the time from nephrectomy to death from any cause, may lead to immortal time bias since this study enrolled patients with recurrent or metastatic RCC and deaths can occur in patients without recurrent or metastatic RCC. However, the immortal time bias in this study may have little impact on the generalisability of these results. For the concordance analysis, we evaluated PD-L1 status using the diagnostic algorithm of SP142, even when the sample was stained with SP263. Since the diagnostic algorithm is specific for each antibody, clinical interpretation of results from this analysis is limited.
In this study, CD8-positive T-cell exhaustion status and presence of other immune related cells were not evaluated. Further investigation is needed for better understanding of the excluded phenotype. TNM staging information was also not obtained because clinical stage was deemed sufficient, considering that the ARCHERY study focused on OS after first line treatment of recurrent or metastatic RCC. Although nuclear grade is a known prognostic marker after radical nephrectomy, it was not considered in this exploratory analysis as a stratification factor since the number of patients was limited.
This exploratory analysis provides insight into the association between two widely known immunological prognostic factors (immune phenotype and PD-L1 expression), improving the understanding of factors that affect RCC prognosis. We observed that immune phenotype was highly associated with PD-L1 expression. Although both covariates were associated with nuclear grade, which is a known prognostic marker in RCC, it is still important to understand the tumour immune environment when predicting prognosis with RCC. Furthermore, concordance analysis of the SP142 and SP263 IHC assays showed a high level of positive percent agreement and moderate levels of overall and negative percent agreement.
The authors would like to thank the patients who participated in the trial, the patients' families, and the investigators, pathologists, and staff at all participating centres (Supplementary Table 6, Appendix A). Third-party writing assistance for this manuscript was provided by Bena Lim, PhD, of Health Interactions, and supported by Chugai Pharmaceutical Co., Ltd.
Conflicts of interest and sources of funding
This study was funded by Chugai Pharmaceutical Co., Ltd. Toyonori Tsuzuki received honoraria from Astellas Pharma, AstraZeneca, Bayer, Bristol Myers Squibb, Chugai Pharmaceutical, Daiichi Sankyo, Janssen Pharmaceuticals, Nippon Kayaku, Novartis, Ono Pharmaceutical, and Takeda; served as an advisor or consultant for AstraZeneca, Bristol Myers Squibb, Chugai Pharmaceutical, and Ono Pharmaceutical; and received research funding from Chugai Pharmaceutical. Chisata Ohe received honoraria from AstraZeneca, Chugai Pharmaceutical, and Janssen Pharmaceuticals; travel, accommodation, or other expenses from AstraZeneca and Chugai Pharmaceutical; and research funding from Chugai Pharmaceutical. Hirotsugu Uemura received honoraria from Astellas Pharma, Bayer, Bristol Myers Squibb, Janssen Pharmaceuticals, Merck Sharp & Dohme, Ono Pharmaceutical, and Pfizer; served as an advisor or consultant for Merck Sharp & Dohme, Ono Pharmaceutical, Pfizer, and Sanofi; and received research funding from AstraZeneca, Astellas Pharma, Daiichi Sankyo, Janssen Pharmaceuticals, Ono Pharmaceutical, Pfizer, Sanofi, Taiho Pharmaceutical, and Takeda. Go Kimura received honoraria from AstraZeneca, Bayer, Bristol Myers Squibb, Chugai Pharmaceutical, Janssen Pharmaceuticals, Merck Sharp & Dohme, Novartis, Ono Pharmaceutical, Pfizer, and Takeda; and research funding from AstraZeneca, Bayer, Bristol Myers Squibb, Chugai Pharmaceutical, Janssen Pharmaceuticals, Merck Sharp & Dohme, Ono Pharmaceutical, and Takeda. Nobuaki Matsubara served as an advisor or consultant for Janssen and Sanofi; and received research funding from Amgen, Astellas Pharma, AstraZeneca, Bayer, Chugai Pharmaceutical, Eisai, Eli Lilly, Janssen Pharmaceuticals, Pfizer, PRA Health Sciences, Taiho Pharmaceutical, and Merck Sharp & Dohme. Takuya Yoshimoto, Yuki Nakagawa, and Tamaki Fukuyama are employees of Chugai Pharmaceutical. Takahiro Osawa, Yosuke Yasuda, Toshiaki Tanaka, Satoshi Anai, Kazutoshi Yamana, and Shingo Hatakeyama declare no conflict of interest.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
T-cell-inflamed gene-expression profile, programmed death ligand 1 expression, and tumor mutational burden predict efficacy in patients treated with pembrolizumab across 20 cancers: KEYNOTE-028.
Efficacy and correlative analyses of avelumab plus axitinib versus sunitinib in sarcomatoid renal cell carcinoma: post hoc analysis of a randomized clinical trial.
Atezolizumab plus bevacizumab versus sunitinib for patients with untreated metastatic renal cell carcinoma and sarcomatoid features: a prespecified subgroup analysis of the IMmotion151 clinical trial.
An outcome prediction model for patients with clear cell renal cell carcinoma treated with radical nephrectomy based on tumor stage, size, grade and necrosis: the SSIGN score.
00259-8&id=gr1.jpg){kind=link}
00259-8&id=gr2.jpg){kind=link}