You are here

Article Text

Article menu
Clinical/translational cancer immunotherapy
Original research
Comparison of efficacy discrepancy between early-phase clinical trials and phase III trials of PD-1/PD-L1 inhibitors
  1. Xiang Li1,2,
  2. Yangzhong Zhou3,
  3. Bing Xu1,
  4. Yunhe Qin4,
  5. Jiuliang Zhao3,
  6. Mengtao Li3,
  7. Jiachen Xu5 and
  8. Guanqiao Li1,6
  1. 1Vanke School of Public Health, Tsinghua University, Beijing, Beijing, China
  2. 2School of Medicine, Tsinghua University, Beijing, Beijing, China
  3. 3Department of Rheumatology, Peking Union Medical College Hospital (PUMCH), Peking Union Medical College & Chinese Academy of Medical Sciences, National Clinical Research Center for Dermatologic and Immunologic Diseases (NCRC-DID), Key Laboratory of Rheumatology and Clinical Immunology, Ministry of Education, Beijing, China, Beijing, China
  4. 4Pharmcube (Beijing) Co Ltd, Beijing, China
  5. 5Department of Medical Oncology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China
  6. 6Institute for Healthy China, Tsinghua University, Beijing, Beijing, China
  1. Correspondence to Dr Guanqiao Li; guanqiaoli{at}tsinghua.edu.cn; Dr Jiachen Xu; xujc{at}cicams.ac.cn

Abstract

Background Phase III clinical trials are pivotal for evaluating therapeutics, yet a concerning failure rate has been documented, particularly impacting oncology where accelerated approvals of immunotherapies are common. These failures are predominantly attributed to a lack of therapeutic efficacy, indicating overestimation of results from phase II studies. Our research aims to systematically assess overestimation in early-phase trials involving programmed cell death-1 (PD-1)/programmed cell death-ligand 1(PD-L1) inhibitors compared with phase III trials and identify contributing factors.

Methods We matched 51 pairs of early-phase and phase III clinical trials from a pool of over 9,600 PD-1/PD-L1 inhibitor trials. The matching criteria included identical treatment regimens, cancer types, treatment lines, and biomarker enrichment strategies. To assess overestimation, we compared the overall response rates (ORR) between early-phase and phase III trials. We established independent variables related to eligibility criteria, and trial design features of participants to analyze the factors influencing the observed discrepancy in efficacy between the two phases through univariable and multivariable logistic analyses.

Result Early-phase trial outcomes systematically overestimated the subsequent phase III results, yielding an odds ratio (OR) comparing ORR in early-phase versus phase III: 1.66 (95% CI: 1.43 to 1.92, p<0.05). This trend of inflated ORR was consistent across trials testing PD-1/PD-L1 monotherapies and combination therapies involving PD-1/PD-L1. Among the examined factors, the exclusion of patients with autoimmune diseases was significantly associated with the disparity in efficacy between early-phase trials and phase III trials (p=0.023). We calculated a Ward statistic of 2.27 to validate the effectiveness of the model.

Conclusion These findings underscore the tendency of overestimation of efficacy in early-phase trials involving immunotherapies. The observed differences could be attributed to variations in the inclusion of patients with autoimmune disorders in early-phase trials. These insights have the potential to inform stakeholders in the future development of cancer immunotherapies.

  • Immunotherapy
  • Clinical Trials as Topic
  • Immune Checkpoint Inhibitors
http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/jitc-2023-007959

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Substantial failure rate in phase III clinical trials, especially in oncology, prompts ethical and resource concerns. This is primarily attributed to a lack of therapeutic efficacy in phase III studies.

    WHAT THIS STUDY ADDS

    • Outcomes of early-phase clinical trials of programmed cell death-1 (PD-1)/programmed cell death-ligand 1(PD-L1) inhibitors have been systematically overestimated compared with subsequent phase III results.

    • Exclusion of patients with autoimmune diseases in early-phase trials has been demonstrated to be the significant determinant of overestimation.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • Our findings provide insights in the design of future clinical trials for cancer immunotherapy. Further investigations are warranted to assess the effectiveness of cancer immunotherapies in patients with pre-existing autoimmune disorders to enhance our understanding and optimize treatment outcomes.

    Introduction

    Phase III clinical trials hold immense importance in assessing the efficacy and safety of emerging therapeutics,1 with positive outcomes serving as a gateway to regulatory approval.2 However, a documented failure rate of approximately 40% in phase III trials has led to substantial resource wastage and ethical concerns, as patients may be exposed to ineffective treatments.3–5 This issue is particularly alarming in the field of oncology, where a substantial proportion of oncology drugs have received accelerated approval based on early-phase clinical data,5 notably observed with programmed cell death-1 (PD-1)/programmed cell death-ligand 1(PD-L1)t inhibitors.6 Immunotherapy has become the mainstream approach in cancer treatment, with thousands of ongoing clinical trials in this field.7 Nonetheless, according to the US Food and Drug Administration, among the 36 accelerated approvals involving PD-1/PD-L1, 8 have already been withdrawn from the market.8

    The primary cause of these failures has been attributed to a lack of therapeutic efficacy, constituting 55% of phase III failures.5 Essentially, this implies the results were overestimated in phase II studies, but could not be reproduced on a larger scale in phase III trials. This pattern of overestimation has been consistently observed in cancer and other indications.9–14 In the context of chemotherapy and targeted therapy in cancer, it has been suggested that factors related to study design and sample size may be implicated, although only preliminary analyses were conducted.15 16 In other indications like rheumatoid arthritis, overestimation has been linked to eligibility criteria of the study population.17

    Currently, there is a notable gap in systematic research within the field of oncology, particularly in the booming PD-1/PD-L1 arena, to identify the precise factors influencing overestimation. Therefore, our research aims to comprehensively assess the extent of overestimation in early-phase trials compared with phase III clinical trials involving PD-1/PD-L1 inhibitors and explore potential factors contributing to this systematic overestimation. This research endeavor is anticipated to yield valuable insights with implications for the design of future clinical trials related to immunotherapies.

    Method

    Search strategy

    We conducted searches on major clinical trial registration platforms including the US’ ClinicalTrials.gov (https://www.clinicaltrials.gov/), the European Union’s EudraCT (https://www.clinicaltrialsregister.eu/), Japan’s UMIN-CTR (https://www.umin.ac.jp/ctr/), Australia’s ANZCTR (https://www.anzctr.org.au/) and China’s CDE (http://www.chinadrugtrials.org.cn/) platforms. To compile a comprehensive data set, we accessed the “trialcube” database of Pharmcube,18 a pharmaceutical data repository that amalgamates studies registered on aforementioned platforms on a daily basis. We identified over 9,600 clinical trials by conducting standardized searches for drugs targeting PD-1/PD-L1 as of February 14, 2023.

    We subsequently narrowed our focus to 3,105 phase III trials, for which we obtained the trial outcomes by searching for the registered numbers of each study across various databases, including PubMed, Web of Science, and scientific conferences related to oncology, such as American Society of Clinical Oncology and European Society for Medical Oncology. The data extraction was completed by April 16, 2023. Among the 852 phase III trials with available outcomes, we further conducted matching with early-phase trials (see below in the “Matching strategy” section). The variables of interest in these trials included trial phase, treatment regimens, indications, lines of therapies, inclusion and exclusion criteria, primary outcomes and efficacy outcomes regarding overall response rates. Majority of the PD-1/PD-L1 products have been approved for certain indications in major countries.

    Matching strategy

    After identifying relevant literature on clinical trials, we deployed a matching process to pair the phase III trials with early-phase trials. Our matching criteria were guided by the following principles: (1) same regimens involving PD-1/PD-L1 inhibitors; (2) same types of cancer, including well-defined molecular types if specified; (3) identical number of lines of treatment, (4) in case where PD-L1 expression or microsatellite instability signature was relevant to the study. These criteria needed to align between at least one study group in the paired trials (eg, both trials involved patients with PD-L1 expression larger than 1%); and (5) the early-phase trials had to be cited as in the published literature of the phase III trial. Following these principles outlined above, we identified and paired 51 sets of clinical trials. The early-phase trials primarily consisted of phase II trials. In cases where phase II trials meeting the criteria could not be found, we extended our search to include phase I trials within the same context. These 51 sets comprised 23 pairs involving monotherapies of PD-1/PD-L1 inhibitors and 28 pairs with combination regimens (figure 1).

    Figure 1
    Figure 1

    Flow chart of study screening, selection, and matching process for programmed cell death-1 (PD-1)/programmed cell death-ligand 1(PD-L1)-related clinical trials.

    Data extraction

    Trial characteristics were retrieved from various sources, including literature, ClinicalTrials.gov, EudraCT, and official trial reports. The primary outcome of efficacy in our study was overall response rates (ORR) in both early-phase and phase III clinical trials. ORR was chosen as it provides an objective measure of drug activity, focusing on the immediate responses without considering long-term effects. It can be evaluated in various single-arm studies, enabling the comparison of single arms between paired trials. Other trial characteristics included sample size, the number of study centers, and the year of publication. In addition, we conducted an extraction of eligibility criteria, encompassing the following factors: life expectancy greater than 3 or 6 months, the requirement for representative tumor specimens, the presence of measurable disease according to the Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 criteria, adequate hematologic function, adequate hepatic function, adequate renal function, PD-L1 tumor cells or immune cells status determined by immunohistochemistry, biomarker expression of targeted cancer, central nervous system disorders (including metastasis), currently active infection, prior chemotherapy treatment, prior immunotherapy treatment, prior immunosuppressive therapy, history of malignancies other than the cancer of interest, history of autoimmune diseases, history of interstitial pulmonary diseases, history of hepatitis virus infection, history of HIV infection, history of cardiovascular diseases, history of organ transplantation, vaccination within the past weeks, pregnancy or lactation status.

    Data analysis

    To explore potential differences in efficacy between early-phase and phase III trials, we first compared the ORR between early-phase studies and corresponding phase III studies using scatter plots. Forest plots were created to visually display the odds ratio (OR) and associated CIs for each trial’s ORR in early-phase trials relative to phase III trials with a random effects model. Furthermore, funnel plots were generated to investigate the relationship between study size and effect size. All analyses were performed individually for each clinical trial. Given that our analysis was conducted at the level of individual study arms, and some study arms originated from the same clinical trials, calculations of heterogeneity were infeasible.

    To analyze the factors that might influence the disparities in efficacy between early-phase and phase III trials, we introduced a dichotomous variable called ORv. ORv served as the dependent variable, representing the CI of the OR for ORR in early-phase trials relative to phase III trials, as calculated during the meta-analysis. This variable was assigned a value of 1 if the CI of OR was entirely to the right of 1, and a value of 0 if the CI had a fraction less than 1.

    For the independent variables, we defined a set of variables related to phase, clinical trial design, and eligibility criteria of the study population, aiming to investigate their impact on the differences in efficacy between early-phase and phase III trials. These variables were defined as binary, with a value of 1 indicating the presence of a particular characteristic or difference between early-phase and corresponding phase III trials, and a value of 0 indicating its absence.

    Finally, we employed a univariable logistic regression approach, analyzing each independent variable separately to assess its significant contribution to the differences in efficacy between early-phase and phase III trials. To ensure the adequacy of our model fit, we used likelihood ratio tests and the Akaike information criterion, prioritizing models that strike an optimal balance between simplicity and explanatory power. We also conducted multivariable logistic regression analysis on variables that demonstrated significance in the univariate analysis, as well as on variables that were closely related. The model’s effectiveness was further evaluated using the Ward statistic. Variance inflation factor was calculated to assess potential issues of collinearity among the variables. To mitigate the risk of false positives, the Benjamini-Hochberg method was applied, enabling us to control the false discovery rate and ensure the reliability of our p value findings. All statistical analyses were executed using RStudio V.4.2.3 and Stata V.17. A two-sided p value <0.05 was considered statistically significant.

    Risk of bias

    In our assessment of included randomized trials, we applied the Cochrane Collaboration’s Risk of Bias Assessment Tool to rigorously evaluate six key types of bias: selection, performance, detection, attrition, reporting, and other potential biases.19 To guarantee the objectivity and precision of our evaluation, each study was independently by two independent researchers. In cases where their assessments differed, the researchers engaged in detailed discussions to achieve a consensus. The outcome of this process was the categorization of each study based on the risk of bias, which was designated as low, high, or unclear for each bias type. An “unclear risk of bias” categorization was applied in situations where the information provided was either insufficient or inadequately detailed, rendering a definitive assessment of bias risk unfeasible.

    Results

    Comparison of outcomes between early-phase and phase III trials

    A visual examination of the ORR of early-phase and phase III trials assessing the identical treatment regimens indicated an overestimation of ORR values in early-phase trials compared with phase III trials, as depicted in figure 2. This trend of inflated ORR in early-phase trials was consistent across trials testing PD-1/PD-L1 monotherapies and combination therapies involving PD-1/PD-L1.

    Figure 2
    Figure 2

    Comparative analysis of overall response rates between early-phase and phase III studies. Green dots represent studies that use monotherapy with programmed cell death-1 (PD-1)/programmed cell death-ligand 1(PD-L1) inhibitors, and red marks denote those involving combination therapies. The size of each dot corresponds to the total number of participants of the two selected arms in both early-phase and phase III studies.

    To quantify this overestimation, the random-effects model analysis reveals an OR of overestimation is 1.67 (95% CI: 1.44 to 1.94, p<0.01, figure 3). Dissecting this further, trials using single agent PD-1/PD-L1 inhibitors showed an OR of 1.45 (95% CI: 1.18 to 1.79, p<0.01), while trials incorporating combination drug regimens demonstrated a more pronounced OR of 1.85 (95% CI: 1.52 to 2.25, p<0.01). Further reinforcing these results, sensitivity analyses, which included stepwise exclusion (refer to online supplemental figure 1) and subgroup analyses (see online supplemental figure 2), consistently yielded an OR greater than 1, substantiating the robustness of our findings.

    Supplemental material

    Supplemental material

    Figure 3
    Figure 3

    Random-effect model analysis for outcome comparison between early-phase and phase III trials involving anti-PD-1/PD-L1 inhibitors. Forest plots visualizing ORs of objective response rates with±95% CIs in early-phase studies compared with phase III studies. Each row presents data for a specific treatment regimen used in a particular population for a specific type of cancer as the unit of analysis. The squares represent the paired evaluation of outcomes in matched early-phase and phase III trials, and the bars (lines) represent the 95% CI for each comparison. Summary estimates are depicted as diamonds at the bottom of the plot, encompassing±95% CI. Results from heterogeneity analysis are displayed beneath the plot. CRC, carcinoma of colon and rectum; EP: Early-phase trials; HCC, hepatic cell cancer; HL, Hodgkin’s lymphoma; HNSCC, head and neck squamous cell carcinoma; MPM, malignant pleural mesothelioma; NSCLC, non-small-cell lung cancer; P3: Phase III trials; PD-1, programmed cell death-1; PD-L1, programmed cell death-ligand; RCC, renal cell cancer; TNBC, triple-negative breast cancer; 1/2/3 L: first/second/third-line therapy.

    Factors contributing to discrepancies in efficacy between early-phase and phase III trials

    To explore the reasons behind the efficacy overestimation in early-phase trials, determinants that contribute to the observed differences between early-phase and phase III trials were investigated (figure 4). Among the selected factors, exclusion of patients with autoimmune diseases stood out as the sole variable significantly influencing the disparity in efficacy (p=0.023). The computed Wald statistic was 2.27 for the exclusion criteria of autoimmune diseases, indicating a significant deviation from the null hypothesis at the 0.05 level and substantiating its role in the overestimation of efficacy in early-phase trials compared with phase III trials. Further strengthening our findings, we conducted multivariable logistic analysis incorporating several variables including the exclusion criteria of autoimmune diseases (online supplemental table 4). These analyses consistently demonstrate that the exclusion of patients with pre-existing autoimmune diseases markedly influence the overestimation of efficacy in these trials.

    Supplemental material

    Figure 4
    Figure 4

    Exploratory analysis of efficacy differences between early-phase and phase III trials involving PD-1/PD-L1 immunotherapies. (A) Heat map visualizing the association of the evaluated variables and the observed efficacy differences. Color coding and numeric indicators within each cell represent the p values of each determinant, as determined through univariable analysis. This analysis tests the null hypothesis (H0) that the CI of OR lies entirely to the right of 1. (B) Scatter plot to investigate the impact of the inclusion or exclusion of patients with autoimmune diseases in the eligibility criteria on the overestimation of efficacy in early-phase trials versus phase III trials. Each dot indicates one pair of trials, plotting the OR for overall response rates in early-phase compared with phase III on the y-axis, against the eligibility criteria of patients with autoimmune diseases on the x-axis. The labels on the x-axis are marked with (+) to denote “trials that included an exclusion criterion for patients with autoimmune diseases” and (–) to indicate the absence of such a criterion. CNS: central nervous system; ECOG: Eastern Cooperative Oncology Group Performance Status; IHC, immunohistochemistry; PD-1, programmed cell death-1; PD-L1, programmed cell death-ligand; RECIST: Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 criteria.

    To further delineate this correlation, we stratified the trial pairs based on their inclusion and exclusion criteria regarding autoimmune diseases in both early-phase and phase III studies. The analysis revealed that trials which excluded patients with autoimmune diseases in the early phase but not in phase III exhibited a significantly elevated OR for efficacy discrepancy when contrasted with trials that consistently applied autoimmune disease exclusion criteria across both phases.

    Risk of bias

    Individual study outcomes (online supplemental table 1) and risk of bias assessments (online supplemental table 2) are listed in the supplementary tables. A majority of the studies analyzed were assigned a low risk of bias. The most frequently encountered issue contributing to an unclear risk of bias designation stemmed from insufficient information regarding random-sequence generation. None of the studies included in our analysis were classified as having a high risk of bias, affirming the methodological rigor of the research underpinning our findings.

    Discussion

    This study demonstrates that early-phase clinical trials for PD-1/PD-L1 inhibitors in tumor treatment tend to yield higher efficacy estimates compared with phase III trials. Our analysis indicates an OR of 1.67 (95% CI: 1.44 to 1.94, p<0.01) for the overestimation of ORR in these early-phase trials. This pattern of early-phase results not alighting with phase III outcomes is not exclusive to oncology, but is also observable in various other diseases,9–14 16 17 20–22 including autoimmune diseases (such as rheumatoid arthritis, psoriatic arthritis and Crohn’s disease, multiple sclerosis, and systemic lupus erythematosus) and certain malignancies (like pancreatic cancer, and cutaneous T-cell lymphoma). Our findings, therefore, add to a growing body of evidence suggesting caution in interpreting early-phase trial results as definitive indicators of phase III trial outcomes. However, the reason for this inconsistency was less known.

    Our findings indicate that excluding patients with autoimmune disorders was significantly associated with the discrepancy of effect estimation, in contrast to other criteria which show minimal impact. Specifically, the ORR in early-phase trials was significantly higher than the paired phase III trials, when patients with autoimmune disorders were excluded in early-phase trials but not in phase III trials. However, other variables, including inclusion criteria, and baseline characteristics, on the difference in efficacy between early-phase and phase III trials, did not demonstrate statistically significant impact on the efficacy variance. Given the larger sample size of phase III studies, these observed differences likely stem from an overestimation of treatment efficacy in early-phase studies, rather than an underestimation in phase III data.

    The interplay between autoimmune diseases and cancer risk is multifaceted, with the risk profile varying among different autoimmune disorders. While a generalized increase in cancer risk is observed, the extent and nature of this risk differ across various conditions.23 For instance, patients with systemic lupus erythematosus have an increased risk of developing solid tumors like bladder, endometrial, and cervical cancers, as well as hematological malignancies. Additionally, inflammatory myopathies, within 3 years of diagnosis, reveal a 74.6% likelihood of progressing to malignancies, with a pronounced risk for ovarian cancer.23 24 In addition, many drugs used to treat autoimmune diseases have carcinogenic properties. Cyclophosphamide has been classified as a Group 1 carcinogen by the WHO.25 Despite of these, it remains to be thoroughly investigated whether the underlying autoimmune condition alters the characteristics of tumors in a manner that significantly impacts the efficacy of pharmacological treatments.

    The integration of tumor immunotherapy with treatment for autoimmune diseases in patients with cancer presents a complex challenge, particularly regarding the safety of immunotherapy. Concerns focus on the potential for recurrence of autoimmune diseases and emergence of immune-related adverse events, which may be severe. Consequently, patients with autoimmune diseases are often excluded from clinical trials. Recent evidence indicated that immune checkpoint inhibitors are safe for well-managed autoimmune diseases, such as rheumatoid arthritis.26 Despite an increased overall risk of immune-related adverse reactions, the incidence of severe adverse events and mortality risks does not appear to be elevated.

    Moreover, the antitumor efficacy of immunotherapy for patients with autoimmune diseases poses another challenge. In our study, early-phase trials overestimate the efficacy of immunotherapy, potentially skewed by excluding autoimmune disease patients. This suggests that the efficacy might be lower in the recruited patients with previously autoimmune diseases in paired phase III trials. Unfortunately, the lack of baseline data on immune diseases in these studies hampers further validation of this finding. Literature review indicate that immune checkpoint inhibitors in patients with a history of autoimmune disease can trigger relapses and necessitate immunosuppressive therapy, thereby affecting their efficacy.27–29 Moreover, the efficacy of PD-1/PD-L1 inhibitors in these patients may be reduced due to the pre-existing immune system dysregulation. Altered immune state, characterized by prolonged hyperactivity and exposure to interferons, has been featured in these patients.30 31 Downregulated PD-1 expression in both CD4+ and CD8+ T cells,32 33 as well as increased levels of anti-PD-1 antibodies,34 has been detected and potentially undermines the effectiveness of PD-1/PD-L1 inhibitor therapies in treating tumors. While some studies reported comparable efficacy in autoimmune and non-autoimmune patients,35 36 these were limited to the use of ipilimumab in melanoma, and could not fully address our concerns. Further meticulously designed trials are needed to directly validate the response of these patients to immunotherapy.

    To enhance the validity of our results in assessing the efficacy of PD-1/PD-L1 inhibitors, we employed rigorous sensitivity analyses. A stepwise exclusion method was applied to each study, confirming the stability of the combined OR and the robustness of our meta-analysis findings (online supplemental figure 1). Subgroup analysis, considering factors like treatment regimen, PD-L1 expression, treatment stages, cancer types, and study characteristics, consistently yielded OR values greater than 1, aligning with our primary conclusion (online supplemental figure 2).

    Our study has several limitations. First, our independent variables were all dichotomous, neglecting the details of some quantitative variables, such as participant number, blood count and liver function measures, and PD-L1 expression. The latter is particularly important but cannot be quantified due to the inconsistent descriptions of PD-L1 expression level in each clinical trial, thus limiting our ability to assess its impact on efficacy differences. Some important factors, though recognized as potential confounders in literature, were unable to be analyzed due to lack of data, such as concurrent antibiotic utilization. Second, the heterogeneity among trials and potential publication bias were noted as concerns. Given the lack of data availability, we were not be able to precisely quantify the heterogeneity and publication bias. Third, possibility for experiment-wise error, especially in early phase trials with smaller sample sizes, could not be neglected. Smaller sample size may make the results susceptible to random errors, leading to unstable or biased outcomes. Fourth, ORR was used as an integrated efficacy endpoint in our analysis, due to its feasibility to be retrieved from both early-phase trials and phase III trials. Its limitations in reflecting survival outcomes were acknowledged, and it is rarely considered as the primary endpoint in phase III trials. Finding better indicators to replace ORR and comparing the differences in survival outcomes between early-clinical trials and phase III clinical trials for immunotherapies still pose a significant challenge.

    In conclusion, our systematic literature review suggests a possible overestimation of efficacy in early-phase cancer immunotherapy trials, partly attributable to the exclusion of autoimmune patients. We analyzed 51 pairs of early-phase and phase III clinical trials, the largest cohort to date, focusing on PD-1/PD-L1 as a therapeutic agent across diverse cancer types. This underscores the importance of our findings in guiding future development in immunotherapy. Due to the concerns regarding potential diminished efficacy, inclusion of patients with autoimmune diseases from trials raises uncertainties. We recommend designing future trials specifically for these patients to deepen our understanding and improve treatment outcomes.

    Ethics statements

    Patient consent for publication

    Ethics approval

    Not applicable.

    Acknowledgments

    We would like to express our gratitude to Qixian Yang (Pharmcube) for his assistance in data standardization and matching.

    References

    1. ↵
      1. Dent SF,
      2. Eisenhauer EA
      . Phase I trial design: are new Methodologies being put into practice Ann Oncol 1996;7:561–6. doi:10.1093/oxfordjournals.annonc.a010671
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Yan F,
      2. Thall PF,
      3. Lu KH, et al
      . Phase I-II clinical trial design: a state-of-the-art paradigm for dose finding. Ann Oncol 2018;29:694–9. doi:10.1093/annonc/mdx795
      OpenUrl
    3. ↵
      1. Dowden H,
      2. Munro J
      . Trends in clinical success rates and therapeutic focus. Nat Rev Drug Discov 2019;18:495–6. doi:10.1038/d41573-019-00074-z
      OpenUrlCrossRefPubMed
    4. ↵
      1. Hay M,
      2. Thomas DW,
      3. Craighead JL, et al
      . Clinical development success rates for investigational drugs. Nat Biotechnol 2014;32:40–51. doi:10.1038/nbt.2786
      OpenUrlCrossRefPubMed
    5. ↵
      1. Harrison RK
      . Phase II and phase III failures: 2013-2015. Nat Rev Drug Discov 2016;15:817–8. doi:10.1038/nrd.2016.184
      OpenUrlCrossRefPubMed
    6. ↵
      1. Li G,
      2. Jiang Y,
      3. Qin Y, et al
      . Comparing development strategies for Pd1/Pdl1-based Immunotherapies. Nat Rev Drug Discov 2022;21:484. doi:10.1038/d41573-022-00003-7
    7. ↵
      1. Saez-Ibañez AR,
      2. Sommers E,
      3. Upadhaya S, et al
      . Pd1/Pdl1 clinical trials adapt to a growing landscape of patients resistant to Pdx. Nat Rev Drug Discov 2023;22:944–5. doi:10.1038/d41573-023-00179-6
      OpenUrl
    8. ↵
    9. ↵
      1. Genovese MC,
      2. Lee E,
      3. Satterwhite J, et al
      . A phase 2 dose-ranging study of subcutaneous Tabalumab for the treatment of patients with active rheumatoid arthritis and an inadequate response to methotrexate. Ann Rheum Dis 2013;72:1453–60. doi:10.1136/annrheumdis-2012-202864
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Genovese MC,
      2. van der Heijde DM,
      3. Keystone EC, et al
      . A phase III, multicenter, randomized, double-blind, placebo-controlled, parallel-group study of 2 dosing regimens of Fostamatinib in patients with rheumatoid arthritis with an inadequate response to a tumor necrosis factor-alpha antagonist. J Rheumatol 2014;41:2120–8. doi:10.3899/jrheum.140238
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Breneman D,
      2. Duvic M,
      3. Kuzel T, et al
      . Phase 1 and 2 trial of Bexarotene GEL for skin-directed treatment of patients with cutaneous T-cell lymphoma. Arch Dermatol 2002;138:325–32. doi:10.1001/archderm.138.3.325
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Heald P,
      2. Mehlmauer M,
      3. Martin AG, et al
      . Topical Bexarotene therapy for patients with refractory or persistent early-stage cutaneous T-cell lymphoma: results of the phase III clinical trial. J Am Acad Dermatol 2003;49:801–15. doi:10.1016/s0190-9622(03)01475-0
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Comi G,
      2. Pulizzi A,
      3. Rovaris M, et al
      . Effect of Laquinimod on MRI-monitored disease activity in patients with relapsing-remitting multiple sclerosis: a Multicentre, randomised, double-blind, placebo-controlled phase IIb study. Lancet 2008;371:2085–92. doi:10.1016/S0140-6736(08)60918-6
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Vollmer TL,
      2. Sorensen PS,
      3. Selmaj K, et al
      . A randomized placebo-controlled phase III trial of oral Laquinimod for multiple sclerosis. J Neurol 2014;261:773–83. doi:10.1007/s00415-014-7264-4
      OpenUrlCrossRefPubMed
    15. ↵
      1. El-Maraghi RH,
      2. Eisenhauer EA
      . Review of phase II trial designs used in studies of molecular targeted agents: outcomes and predictors of success in phase III. J Clin Oncol 2008;26:1346–54. doi:10.1200/JCO.2007.13.5913
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Zia MI,
      2. Siu LL,
      3. Pond GR, et al
      . Comparison of outcomes of phase II studies and subsequent randomized control studies using identical chemotherapeutic regimens. J Clin Oncol 2005;23:6982–91. doi:10.1200/JCO.2005.06.679
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Kerschbaumer A,
      2. Smolen JS,
      3. Herkner H, et al
      . Efficacy outcomes in phase 2 and phase 3 randomized controlled trials in rheumatology. Nat Med 2020;26:974–80. doi:10.1038/s41591-020-0833-4
      OpenUrlPubMed
    18. ↵
      1. Li G,
      2. Qin Y,
      3. Xie C, et al
      . Trends in oncology drug innovation in China. Nat Rev Drug Discov 2021;20:15–6. doi:10.1038/d41573-020-00195-w
      OpenUrl
    19. ↵
      1. Higgins JPT,
      2. Altman DG,
      3. Gøtzsche PC, et al
      . The Cochrane collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011;343:d5928. doi:10.1136/bmj.d5928
    20. ↵
      1. Sandborn WJ,
      2. Schreiber S,
      3. Feagan BG, et al
      . Certolizumab Pegol for active Crohn’s disease: a placebo-controlled, randomized trial. Clinical Gastroenterology and Hepatology 2011;9:670–678. doi:10.1016/j.cgh.2011.04.031
      OpenUrl
    21. ↵
      1. Schreiber S,
      2. Rutgeerts P,
      3. Fedorak RN, et al
      . A randomized, placebo-controlled trial of Certolizumab Pegol (Cdp870) for treatment of Crohn’s disease. Gastroenterology 2005;129:807–18. doi:10.1053/j.gastro.2005.06.064
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Smolen JS,
      2. Weinblatt ME,
      3. van der Heijde D, et al
      . Efficacy and safety of Tabalumab, an anti-B-cell-activating factor Monoclonal antibody, in patients with rheumatoid arthritis who had an inadequate response to methotrexate therapy: results from a phase III Multicentre, randomised, double-blind study. Ann Rheum Dis 2015;74:1567–70. doi:10.1136/annrheumdis-2014-207090
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. He MM,
      2. Lo CH,
      3. Wang K, et al
      . Immune-mediated diseases associated with cancer risks. JAMA Oncol 2022;8:209–19. doi:10.1001/jamaoncol.2021.5680
      OpenUrl
    24. ↵
      1. Zhou Z,
      2. Liu H,
      3. Yang Y, et al
      . The five major autoimmune diseases increase the risk of cancer: Epidemiological data from a large-scale cohort study in China. Cancer Communications 2022;42:435–46. doi:10.1002/cac2.12283 Available: https://onlinelibrary.wiley.com/toc/25233548/42/5
      OpenUrl
    25. ↵
      1. Mostafa M,
      2. Chalvardjian K,
      3. Khairy A, et al
      . In-vivo and in-vitro evaluations for the effect of cyclophosphamide on the carcinogen Metabolizing system of Mouse-liver microsomes. Oncol Rep 1994;1:249–53. doi:10.3892/or.1.1.249
      OpenUrlPubMed
    26. ↵
      1. McCarter KR,
      2. Wolfgang T,
      3. Arabelovic S, et al
      . Mortality and immune-related adverse events after immune Checkpoint inhibitor initiation for cancer among patients with pre-existing rheumatoid arthritis: a retrospective, comparative, cohort study. Lancet Rheumatol 2023;5:e274–83. doi:10.1016/s2665-9913(23)00064-4
      OpenUrl
    27. ↵
      1. Tison A,
      2. Quéré G,
      3. Misery L, et al
      . Safety and efficacy of immune Checkpoint inhibitors in patients with cancer and preexisting autoimmune disease: A nationwide, multicenter cohort study. Arthritis Rheumatol 2019;71:2100–11. doi:10.1002/art.41068
      OpenUrl
    28. ↵
      1. Hoa S,
      2. Laaouad L,
      3. Roberts J, et al
      . Preexisting autoimmune disease and immune-related adverse events associated with anti-PD-1 cancer Immunotherapy: a national case series from the Canadian research group of rheumatology in Immuno-oncology. Cancer Immunol Immunother 2021;70:2197–207. doi:10.1007/s00262-021-02851-5
      OpenUrl
    29. ↵
      1. Xie W,
      2. Huang H,
      3. Xiao S, et al
      . Immune Checkpoint inhibitors therapies in patients with cancer and preexisting autoimmune diseases: A meta-analysis of observational studies. Autoimmun Rev 2020;19:102687. doi:10.1016/j.autrev.2020.102687
      OpenUrl
    30. ↵
      1. Banchereau J,
      2. Pascual V
      . Type I interferon in systemic lupus erythematosus and other autoimmune diseases. Immunity 2006;25:383–92. doi:10.1016/j.immuni.2006.08.010
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      1. Ramaswamy M,
      2. Tummala R,
      3. Streicher K, et al
      . The pathogenesis, molecular mechanisms, and therapeutic potential of the interferon pathway in systemic lupus erythematosus and other autoimmune diseases. Int J Mol Sci 2021;22:11286. doi:10.3390/ijms222011286
    32. ↵
      1. Kristjansdottir H,
      2. Steinsson K,
      3. Gunnarsson I, et al
      . Lower expression levels of the programmed death 1 receptor on Cd4+Cd25+ T cells and correlation with the PD-1.3A genotype in patients with systemic lupus erythematosus. Arthritis Rheum 2010;62:1702–11. doi:10.1002/art.27417
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. McKinney EF,
      2. Lee JC,
      3. Jayne DRW, et al
      . T-cell exhaustion, Co-stimulation and clinical outcome in Autoimmunity and infection. Nature 2015;523:612–6. doi:10.1038/nature14468
      OpenUrlCrossRefPubMed
    34. ↵
      1. Shi H,
      2. Ye J,
      3. Teng J, et al
      . Elevated serum Autoantibodies against Co-inhibitory PD-1 facilitate T cell proliferation and correlate with disease activity in new-onset systemic lupus erythematosus patients. Arthritis Res Ther 2017;19:52. doi:10.1186/s13075-017-1258-4
    35. ↵
      1. Johnson DB,
      2. Sullivan RJ,
      3. Ott PA, et al
      . Ipilimumab therapy in patients with advanced Melanoma and preexisting autoimmune disorders. JAMA Oncol 2016;2:234–40. doi:10.1001/jamaoncol.2015.4368
      OpenUrl
    36. ↵
      1. Gutzmer R,
      2. Koop A,
      3. Meier F, et al
      . Programmed cell death Protein-1 (PD-1) inhibitor therapy in patients with advanced Melanoma and preexisting Autoimmunity or Ipilimumab-triggered Autoimmunity. Eur J Cancer 2017;75:24–32. doi:10.1016/j.ejca.2016.12.038
      OpenUrlPubMed

    Supplementary materials

    Footnotes

    • Contributors Conception and design: XL and GL. Development of methodology: XL, BX and GL. Acquisition of data: XL, YQ and GL. Analysis and interpretation of data: XL, YZ and BX. Drafting of the manuscript: XL, YZ and GL. Review and/or revision of the manuscript: XL, YZ, JX, BX, JZ, ML and GL. Administrative, technical, or material support: JX, JZ, ML and GL. Study supervision: GL. Final approval of manuscript: All authors. GL accepted full responsibility for the work, had access to the data, and controlled the decision to publish. GL was responsible for overall content as the guarantor.

    • Funding This work was funded by National Natural Science Foundation of China (72374119), Beijing Nova Program (20231150008), Tsinghua University lnitiative Scientifc Research Program of Precision Medicine (2022ZLA001), Guangdong Association of Clinical Trials Guangdong Association of Clinical Trials (GACT) /Chinese Thoracic Oncology Group (CTONG) (CTONG-YC20220115).

    • Competing interests No, there are no competing interests.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.