Factors Modifying the Associations of Single or Combination Programmed Cell Death 1 and Programmed Cell Death Ligand 1 Inhibitor Therapies With Survival Outcomes in Patients With Metastatic Clear Cell Renal Cell Carcinoma: A Systematic Review and Meta-analysis.

Importance: Programmed cell death 1/programmed cell death ligand 1 (PD-1/PD-L1) inhibitors are immune checkpoint inhibitors widely used in the treatment of metastatic clear cell renal cell carcinoma (ccRCC) and other cancers. There is a lack of understanding regarding which factors are associated with therapeutic response.
Objectives: To conduct a systematic literature review of trials reporting on factors associated with differential response to PD-1/PD-L1 inhibitors among patients diagnosed with metastatic ccRCC and quantitatively synthesize the magnitude to which each factor modified the response to PD-1/PD-L1 inhibitors. Data Sources: The MEDLINE and Cochrane Register of Trials databases were searched for studies published in English from 2006 onward. Searches were last run on September 3, 2019. Study Selection: This systematic review and meta-analysis assessed 662 phase 2/3 randomized clinical trials that provided subgroup analyses of any baseline characteristics regarding the treatment response to PD-1/PD-L1 inhibitors, alone or as part of a combination therapy, with respect to overall survival (OS) or progression-free survival (PFS) among patients with metastatic ccRCC. Data Extraction and Synthesis: A novel quantitative approach was used to synthesize subgroup findings across trials. The ratio of the subgroup-specific hazard ratios (HRs) from each study were pooled using a random-effects meta-analysis whereby ratios of 1.00 would indicate that the subgroup-specific HRs were equal in magnitude. Main Outcomes and Measures: Main outcomes were OS and PFS.
Results: From an initial 662 reports, 7 trials were considered eligible for inclusion. Meta-analyses suggested the treatment response to PD-1/PD-L1 inhibitors in patients with metastatic ccRCC was significantly associated with age (OS: ratio of HR for age ≥75 years to HR for age <65 years, 1.51; 95% CI, 1.01-2.26), PD-L1 expression (PFS: ratio of HR for PD-L1 < 1% to HR for PD-L1 ≥ 10%, 2.21; 95% CI, 1.14-4.27; ratio of HR for PD-L1 < 1% to HR for PD-L1 ≥ 1%, 1.36; 95% CI, 1.10-1.68), Memorial Sloan Kettering Cancer Center risk score (PFS: ratio of HR for immediate risk score to HR for poor risk score, 1.62; 95% CI, 1.14-2.29; ratio of HR for favorable risk score to HR for poor risk score, 1.53; 95% CI, 1.00-2.34; ratio of HR for favorable risk score to HR for intermediate risk score, 0.96; 95% CI, 0.70-1.30), and sarcomatoid tumor presence (PFS: ratio of HR for no sarcomatoid differentiation to HR for sarcomatoid differentiation, 1.54; 95% CI, 1.07-2.21). Conclusions and Relevance: This analysis suggests that older age, low levels of PD-L1 expression, and the absence of sarcomatoid tumor differentiation are associated with a diminished response to anti-PD-1/PD-L1 immunotherapies with respect to survival outcomes among patients with metastatic ccRCC.

Introduction

Although kidney cancer is a relatively less common form of cancer, more than 400 000 new kidney cancer cases occurred across the world in 2018.[1] A global increase in kidney cancer incidence can likely be attributed to improved diagnostic systems, allowing for earlier disease detection.[2] The most common form of kidney cancer is renal cell carcinoma (RCC).[2] The 5-year survival probability for patients with metastatic clear cell RCC (ccRCC) appears to be rapidly increasing. For example, the median overall survival (OS) in the Checkmate 214 (Nivolumab Combined With Ipilimumab Versus Sunitinib in Previously Untreated Advanced or Metastatic Renal Cell Carcinoma) trial is now 47 months, whereas in previous trials, it had ranged from 7 to 29 months.[3,4] Before the introduction of immune checkpoint inhibitors (ICIs), tyrosine kinase inhibitors and mammalian target of rapamycin inhibitors were the standards of care for patients with metastatic ccRCC.[5,6,7,8,9,10] In 2006, sunitinib was granted approval by the US Food and Drug Administration for first-line treatment in metastatic RCC.[5] Shortly thereafter, various trials (ie, COMPARZ [Pazopanib Versus Sunitinib in the Treatment of Locally Advanced and/or Metastatic Renal Cell Carcinoma], Checkmate 025 [Study of Nivolumab Versus Everolimus in Pre-Treated Advanced or Metastatic Clear-Cell Renal Cell Carcinoma], METEOR [A Study Measuring Effects on Intima Media Thickness: An Evaluation of Rosuvastatin 40 mg], CABOSUN [Cabozantinib or Sunitinib Malate in Treating Participants With Metastatic Variant Histology Renal Cell Carcinoma], and Checkmate 214) demonstrated superior efficaciousness and improved safety profiles of ICIs over tyrosine kinase inhibitors and mammalian target of rapamycin therapies.[5,6,7,8,9,10,11,12,13,14,15,16] As such, the treatment landscape for metastatic ccRCC has seen rapid development, with ICIs now occupying earlier treatment lines.

Programmed cell death 1 (PD-1) is a receptor expressed on activated T cells, and programmed cell death ligand 1 (PD-L1) is expressed on dendritic cells.[17,18] Under normal conditions, the binding of these immune checkpoint proteins plays an important physiologic role in minimizing tissue damage by way of controlling inflammation.[18] The PD-L1 can also be expressed on tumor cells; PD-L1–positive tumor cells can use this pathway to escape immune response and further disease progression.[19,20,21]

Monoclonal antibodies have been developed to address this issue of adaptive immune resistance.[20] Anti–PD-1/PD-L1 drugs are types of ICIs that block the signaling pathway between PD-1 and PD-L1, thereby releasing T-cell inhibition and enhancing antitumor activity.[20] Pembrolizumab and nivolumab are examples of anti–PD-1 drugs; avelumab, atezolizumab, and durvalumab are examples of anti–PD-L1 drugs.[20] However, large interpatient variability exists in treatment efficacy and safety and resistance to blockade therapy.[20,21] Identifying which characteristics are associated with response to PD-1/PD-L1 inhibitors is therefore an important task. Specifically, the targeting of ICIs to patients who are most likely to respond to these therapies could lead to cost savings.[22,23]

Research on the effect modification of ICIs is limited. A literature review[24] published in 2017 on the safety and efficacy of ICI use for urologic cancers concluded that more research was needed to determine the association of PD-L1 with anti–PD-1/PD-L1 treatment efficacy. A quantitative synthesis of this evidence base could better assess the association of PD-L1 with ICI therapy response. A recent systematic review and meta-analysis[25] demonstrated that anti–PD-1/PD-L1 inhibitors were associated with survival outcomes in advanced and metastatic cancers compared with conventional therapies. Although the authors quantified the effect of various patient characteristics that might benefit from treatment, their analysis included more studies on lung cancer than any other cancer.

To our knowledge, no review has been conducted on factors associated with differential response to ICIs in metastatic ccRCC. To address this gap, we conducted a systematic review and meta-analysis of subgroup findings from phase 2/3 randomized clinical trials to determine which baseline factors are associated with response to anti–PD-1/PD-L1 treatment in patients with metastatic ccRCC, with respect to survival outcomes. The specific objectives of this study were to systematically examine factors reported in clinical trials that could modify the clinical response to PD-1/PD-L1 inhibitors among patients diagnosed with metastatic ccRCC and to quantitatively synthesize the magnitude of the association of these factors with the treatment response to a PD-1/PD-L1 inhibitor.

Methods

Study Design and Search Strategy

Studies published in English from 2006 onward within the MEDLINE and Cochrane Register of Trials databases were systematically searched using a query that was developed with support from a research librarian at McMaster University (eTables 1 and 2 in the Supplement). The database searches were run on September ‎3, ‎2019. Titles and abstracts were first screened for relevance, and then full-text screening was conducted. Screening was performed in duplicate by 2 investigators (N.S. and S.B.C.). In addition to the database search, reference lists of included articles were also scanned for relevant records. The study was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines.[26]

Study Eligibility

We included all phase 2 or 3 randomized clinical trials that provided subgroup analyses of any baseline characteristics with respect to the effect of a single or combination anti–PD-1/PD-L1 inhibitor on overall survival (OS) or progression-free survival (PFS) among patients with metastatic ccRCC. Studies of patients with early-stage RCC (stage I-III), animal studies, studies that did not use anti–PD-1/PD-L1 inhibitors (ie, atezolizumab, nivolumab, avelumab, pembrolizumab, and durvalumab) alone or in combination with other therapies, studies not conducted within a clinical trial setting, studies that did not conduct a subgroup or regression analysis examining modification by a baseline variable, studies that did not report on OS or PFS, and studies conducted before 2006 were excluded. The inclusion and exclusion criteria are given in eTable 3 in the Supplement. Quality of included studies was assessed through the Cochrane risk-of-bias tool for randomized clinical trials (eFigures 1-29 in the Supplement).[27]

Statistical Analysis

Meta-analyses were conducted using random-effects models to account for heterogeneity.[28] The estimation of the random-effects model was performed using restricted maximum likelihood. The outcome of interest was the ratio of the subgroup-specific hazard ratios (HRs) whereby a value of 1.00 would indicate that the HRs of the 2 subgroups are equal in magnitude and values greater than or less than 1.00 would indicate that the response to treatment was greater or lesser in 1 of the subgroups. The ratio of the subgroup-specific HRs was log transformed before analysis as follows: log(HRA/HRB) = log(HRA) – log(HRB). Because the subgroups within each trial were independent, the variance of the log of the ratio of the subgroup-specific HRs was estimated as follows: var(log[HRA/HRB]) = var(log[HRA]) + var(log[HRB]). The SE of the log of the subgroup-specific HRs was estimated using the reported 95% CIs as follows: SE(log[HR]) = (log[upper confidence limit] – log[lower confidence limit])/3.92.[29] In situations where the 95% CI was only reported in graphical format, the 95% CIs were derived using WebPlotDigitizer, version 4.2.[30] In the meta-analysis, the degree of heterogeneity was quantified using the I2 statistic, a χ2 test, and the 95% prediction interval (PI).[31] Sensitivity analyses were performed using fixed-effects models.[32] Analyses were conducted using R software, version 3.5.0 (R Foundation for Statistical Computing) and the metafor package in R.[33] A 2-sided P < .05 was considered statistically significant.

Results

Included Trials and Studies

The search yielded 662 initial results, from which a total of 9 publications representing 7 unique trials were included in the final review (Figure 1).[34,35,36,37,38,39,40] Six trials[35,36,37,38,39,40] were included in the quantitative meta-analyses; 1 trial[34] was excluded because OS or PFS outcomes were not reported in terms of HRs for any baseline factor. The Table[34,35,36,37,38,39,40] lists the characteristics of the trials included in this review.

Figure 1.

PRISMA Flow Diagram of the Selection Process for Included Studies

Table.

Descriptive Characteristics of the 7 Included Trials

Trial	Phase	Period of enrollment	PD-1/PD-L1 inhibitor(s)	Comparator	Reported outcome(s)	Follow-up time, y	Subgroups examined
NCT01354431 (Motzer et al,[34] May 2015)a	2	May 2011-Jun 2020	Nivolumab	NA	OS, PFS	2	PD-L1
IMmotion150 (McDermott et al,[35] 2018)	2	Jan 2014-Oct 2016	Atezolizumab plus bevacizumab; atezolizumab	Sunitinib	PFS	2.75	IMDC risk score, liver metastases, MSKCC risk score, PD-L1, sarcomatoid presence
Checkmate 025 (Motzer et al,[36] Nov 2015)	3	Sep 2012-May 2015	Nivolumab	Everolimus	OS, PFS	2.5	Age, bone metastases, IMDC risk score, liver metastases, MSKCC risk score, PD-L1, region, sex
Checkmate 214 (Motzer et al,[37] 2018)	3	Oct 2014-Jun 2017	Nivolumab plus ipilimumab	Sunitinib	OS, PFS	2.6	Age, bone metastases, IMDC risk score, lung metastase, PD-L1, region, sex
IMmotion151 (Rini et al,[38] May 2019)	3	May 2015-Sep 2017	Atezolizumab plus bevacizumab	Sunitinib	OS, PFS	2.25 OS; 2 PFS	IMDC risk score, liver metastases, MSKCC risk score, PD-L1, sarcomatoid presence
KEYNOTE-426 (Rini et al,[39] Feb 2019)	3	Sep 2016-Jan 2022	Pembrolizumab plus axitinib	Sunitinib	OS, PFS	3.25 OS; 2 PFS	Age, IMDC risk score, PD-L1, region, sex
JAVELIN Renal 101 (Motzer et al,[40] 2019)	3	Mar 2016-May 2024	Avelumab plus axitinib	Sunitinib	PFS	3.3	Age, IMDC risk score, MSKCC risk score, PD-L1, region, sex

Abbreviations: Checkmate 025, study of Nivolumab Versus Everolimus in Pre-Treated Advanced or Metastatic Clear-Cell Renal Cell Carcinoma; Checkmate 214, Nivolumab Combined With Ipilimumab Versus Sunitinib in Previously Untreated Advanced or Metastatic Renal Cell Carcinoma; IMmotion150, A Study of Atezolizumab (an Engineered Anti-Programmed Death-Ligand 1 [PD-L1] Antibody) as Monotherapy or in Combination With Bevacizumab (Avastin) Compared to Sunitinib (Sutent) in Participants With Untreated Advanced Renal Cell Carcinoma; IMmotion151, a study of Atezolizumab in Combination With Bevacizumab Versus Sunitinib in Participants With Untreated Advanced Renal Cell Carcinoma; IMDC, International Metastatic Renal Cell Carcinoma Database Consortium; JAVELIN Renal 101, a study of Avelumab With Axitinib Versus Sunitinib In Advanced Renal Cell Cancer; MSKCC, Memorial Sloan Kettering Cancer Center; NA, not applicable; OS, overall survival; PD-1, programmed cell death 1; PD-L1, programmed cell death ligand 1; PFS, progression-free survival.

NCT01354431 was not included in any quantitative meta-analyses.

Baseline Factors Included in This Review

This review highlights the main findings of interest. There was evidence that the following baseline factors significantly modified treatment response to an ICI on a survival outcome: age (OS), Memorial Sloan Kettering Cancer Center (MSKCC) risk score (PFS), level of PD-L1 expression (PFS), and sarcomatoid differentiation (PFS) (Figure 2). The PD-L1 expression results that could not be quantitatively synthesized are presented in eTable 4 in the Supplement. The following baseline factors were also quantitatively synthesized, although they did not demonstrate evidence of treatment response modification: age (PFS), bone metastases (OS), International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) risk score (OS and PFS), liver metastases (PFS), lung metastases (OS), PD-L1 (OS), prior nephrectomy (PFS), region (PFS and OS), and sex (OS and PFS) (eFigures 2-27 in the Supplement).

Figure 2.

Random-Effects Meta-analyses on the Ratio of Subgroup-Specific Hazard Ratios (HRs) Among Patients With Metastatic Clear Cell Renal Cell Carcinoma Treated With Anti–programmed Cell Death 1 and Programmed Cell Death Ligand 1 (PD-L1) Therapies

PI indicates prediction interval; all other abbreviations are defined in the first footnote to the Table.

aOlder age subgroup of 65 years or older.

bNot statistically significant.

Quantitative Synthesis: Evidence of Differential Response to Treatment

Age (OS)

In Checkmate 025, Checkmate 214, and KEYNOTE-426 (Study to Evaluate the Efficacy and Safety of Pembrolizumab [MK-3475] in Combination With Axitinib Versus Sunitinib Monotherapy in Participants With Renal Cell Carcinoma), patients 75 years or older had a significantly reduced response to anti–PD-1/PD-L1 therapies compared with younger patients (ratio of HR for age ≥75 years to HR for age <65 years, 1.51; 95% CI, 1.01-2.26; 95% PI, 1.01-2.26; I2 = 0%; P = .04).

MSKCC Risk Score (PFS)

In IMmotion150 (A Study of Atezolizumab [an Engineered Anti-Programmed Death-Ligand 1 (PD-L1) Antibody] as Monotherapy or in Combination With Bevacizumab [Avastin] Compared to Sunitinib [Sutent] in Participants With Untreated Advanced Renal Cell Carcinoma), IMmotion151 (A Study of Atezolizumab in Combination With Bevacizumab Versus Sunitinib in Participants With Untreated Advanced Renal Cell Carcinoma), and JAVELIN Renal 101 (A Study of Avelumab With Axitinib Versus Sunitinib In Advanced Renal Cell Cancer), patients in the intermediate MSKCC risk group at baseline had a significantly reduced response to PD-1/PD-L1 inhibitors compared with patients in the poor risk group (ratio of HR for immediate risk to HR for poor risk, 1.62; 95% CI, 1.14-2.29; 95% PI, 1.14-2.29; I2 = 0%; P = .01). Although not statistically significant, a similar finding was observed when comparing the favorable MSKCC subgroup and the poor subgroup (ratio of HR for favorable risk to HR for poor risk, 1.53; 95% CI, 1.00-2.34; 95% PI, 1.00-2.34; I2 = 0%; P = .05). There was no evidence of a differential response to treatment when comparing the favorable and the intermediate subgroups (ratio of HR for favorable risk to HR for intermediate risk, 0.96; 95% CI, 0.70-1.30; 95% PI, 0.70-1.30; I2 = 0%; P = .77).

PD-L1 (PFS)

IMmotion150, IMmotion151, KEYNOTE-426, and JAVELIN Renal 101 found that the response to PD-1/PD-L1 inhibitors was significantly diminished among patients who expressed PD-L1 less than 1% at baseline compared with patients who expressed PD-L1 of 1% or greater (ratio of HR for PD-L1 < 1% to HR for PD-L1 ≥ 1%, 1.36; 95% CI, 1.10-1.68; 95% PI, 1.10-1.68; I2 = 0%; P = .004). In the IMmotion150 and IMmotion151 trials, patients who expressed PD-L1 less than 1% had a diminished treatment response compared with patients who expressed PD-L1 of 10% or greater (ratio of HR for PD-L1 < 1% to HR for PD-L1 ≥ 10%, 2.21; 95% CI, 1.14-4.27; 95% PI, 1.11-4.39; I2 = 2.26%; P = .02).

Sarcomatoid Differentiation (PFS)

On the basis of the results from the IMmotion150 and IMmotion151 trials, there was evidence that patients without sarcomatoid tumor differentiation had poor response to ICIs compared with patients with sarcomatoid tumors (ratio of HR for no sarcomatoid tumor differentiation to HR for sarcomatoid tumor differentiation, 1.54; 95% CI, 1.07-2.21; 95% PI, 1.07-2.21; I2 = 0%; P = .02).

Discussion

Summary of Main Findings

This systematic review and meta-analysis of the ratio of the subgroup-specific HRs suggests that the response to anti–PD-1/PD-L1 inhibitors was diminished for the following patient groups: older adults (≥75 years), patients with low levels of PD-L1 expression (<1%), patients with a favorable or intermediate MSKCC risk score, and patients without sarcomatoid differentiation. Within this investigation, subgroups of patients who may have a diminished response to anti–PD-1/PD-L1 therapies were identified. However, it is important to recognize that such subgroups may still benefit from PD-1/PD-L1 inhibitors. Although the efficacy of these immunotherapies may be diminished, such therapies may still be more efficacious than the prior standard therapies within these subgroups.

Age

Diminished treatment response to anti–PD-1/PD-L1 therapies was demonstrated in adults 75 years or older when compared with adults younger than 65 years in terms of OS. This finding could be explained by greater susceptibility to the toxic effects associated with immunotherapy in older patients than younger patients.[41] Older patients may also have preexisting conditions or comorbidities that hinder the response to anti–PD-1/PD-L1 treatments.[41] In addition, cellular senescence and autoimmunity both increase with age.[42] Senescent cells can promote inflammation by inducing the release of inflammatory cytokines. This state of chronic inflammation furthers cancer progression. Furthermore, aging is associated with reduced T-cell activation.[41,42] A combination of these factors could be the reason behind the diminished anti–PD-1/PD-L1 response in older adults compared with younger adults. Although older patients may have a decreased treatment response to PD-1/PD-L1 inhibitors, they may still benefit from these immunotherapies.

PD-L1 Expression

Patients with lower levels of PD-L1 expression had diminished ICI treatment response across all trials with regard to PFS. Although the difference in OS HRs was not statistically significant, this could be related to the fact that fewer studies were included in the OS analysis. In general, the magnitude of the treatment response modification by PD-L1 expression levels was similar for both OS and PFS outcomes.

These data suggest that tumors with upregulated PD-L1 expression are more sensitive to PD-1/PD-L1 inhibitors. High levels of PD-L1 expression are associated with a worse prognosis,[43,44] but when treated with PD-1/PD-L1 inhibitors, patients with higher levels of PD-L1 expression respond better to therapy.[44,45] Tumors that have an increased dependence on immunosuppression via PD-L1 are more likely to be affected by the inhibition of the PD-1/PD-L1 pathway. Although patients with upregulated PD-L1 expression may be more sensitive to PD-1/PD-L1 treatments, these immunotherapies may still benefit patients with low levels of PD-L1 expression.

Sarcomatoid Differentiation

Treatment response was diminished in patients without sarcomatoid tumors who were treated with PD-1/PD-L1 inhibitors with regard to PFS. Sarcomatoid RCC is a more aggressive form of cancer; it is found in approximately 5% of all RCCs and is associated with a worse prognosis.[46,47] Evidence indicates that sarcomatoid components in sarcomatoid RCC display increased expression of PD-L1.[47] Thus, the same biological mechanisms elucidating the role of PD-L1 expression on PD-1/PD-L1 inhibition can be applied to understand the associations between sarcomatoid tumors and treatment response. Patients with sarcomatoid differentiation likely have an increased treatment response because the ICIs were able to target the high levels of PD-L1, thereby greatly inhibiting the tumor’s ability to promote immunosuppression. Furthermore, these findings are supported by Checkmate 214 and Keynote-427, which both noted that patients with sarcomatoid differentiation responded well to nivolumab plus ipilimumab and pembrolizumab, respectively.[37,48]

MSKCC Risk Score

The IMDC and MSKCC are widely used to classify patients with metastatic RCC into 3 prognostic groups: favorable, intermediate, and poor.[49] The current study found evidence of a differential response to treatment by MSKCC risk score, with patients with intermediate scores having diminished treatment response when compared with patients with poor scores with regard to PFS. The favorable risk score group also had diminished ICI treatment response compared with the poor risk score group, although this result was not significant. Although the analysis of IMDC risk score (PFS and OS) found similar findings to the MSKCC risk score (PFS), there was no significant demonstration of differential response to treatment among any subgroups. Although further research is needed, these findings suggest that the MSKCC risk score may be better suited to targeting the delivery of ICIs relative to the IMDC risk score. These findings align with the Society for Immunotherapy of Cancer consensus statement wherein 76% of subcommittee members believed that the IMDC categories could not be used to guide the delivery of anti–PD-1/tyrosine kinase inhibitor combination therapy.[13]

Areas for Future Research

Future research could replicate these analyses within different cancer sites and treatment settings. In addition, well-conducted observational studies could be incorporated to address the stringent inclusion and exclusion criteria of randomized clinical trial populations and increase the generalizability of these findings.[50] However, the potential to introduce bias through the inclusion of observational studies should be carefully assessed.[51] Furthermore, the small number of trials included in the study limited the robustness of the meta-analyses. These findings can be updated as more randomized clinical trials are conducted. In addition, the reporting of baseline factors could be improved to better identify which variables would be of interest for future research on the identification of differential response to ICIs. Lastly, because these analyses were based on aggregate-level data, there was no adjustment for variables at the patient level. As such, additional individual-level analyses are required to confirm these findings and explore more complex forms of differential response to treatment in which the analyses are stratified by multiple variables.

Strengths and Limitations

This study has strengths and limitations. It used a novel methodologic approach of predicting differential response to treatment across trials in which subgroup-specific HRs, extracted from subgroup analyses, were pooled via a random-effects meta-analytic model. This approach made it possible to account for the lack of precision in the subgroup analyses presented within the individual trials by pooling findings across trials. In addition, all randomized clinical trials included in the analyses were of sound methodologic quality, and there was little to no heterogeneity between trials.

The findings regarding differential treatment response via age should be interpreted with caution because of the underrepresentation of older patients in clinical trials. The underrepresentation of elderly patients in clinical trials that examine immunotherapy is problematic because results may not be generalizable to real-world patient populations.[52] In addition, these analyses assumed that the magnitude of the difference between subgroups with respect to the treatment response would be similar across different ICI regimens. Within the current context, this assumption should be expected to hold approximately because these treatments have a similar biological mechanism. In support of this hypothesis, there was little to no heterogeneity in these meta-analyses. In addition, results from this investigation may suffer from publication bias if there was an underreporting of exploratory subgroup analyses that were not statistically significant. Unfortunately, there was an insufficient sample size to formally assess the presence of publication bias within these analyses. Therefore, these results should be interpreted with caution. Last, there was a small number of trials included in these analyses, and the sample sizes of the individual trials were limited. Therefore, this analysis may have failed to detect a difference in treatment response for some of the other baseline covariates because of a lack of precision arising from small sample sizes.

Conclusions

Although additional research is needed, results from this meta-analysis suggest that the treatment response to anti–PD-1/PD-L1 treatments was diminished among the following subgroups: patients 75 years or older, patients with low levels of PD-L1 expression (<1%), patients with a favorable or intermediate MSKCC risk score, and patients without sarcomatoid differentiation. These findings can be supported by biological evidence related to age-related immunologic changes and tumor evasion of host immunity. Results from this investigation may help to target the delivery of ICIs among patients with metastatic ccRCC.