Frequently asked questions on surrogate endpoints in oncology-opportunities, pitfalls, and the way forward.

Search strategy and selection criteria

Data for this Review were identified by searches of MEDLINE, Current Contents, PubMed, and references from relevant articles using the search terms “surrogate end points”, “progression free survival”, “overall survival” and “quality of life”. Abstracts and reports from meetings were included only when they related directly to previously published work. Only articles published in English between 2000 and 2023 were included.”

Introduction

Patients with cancer receive various treatments in the hopes of living a longer and better life. The added therapeutic benefit of new treatments versus existing (or no) treatment can be best assessed through randomized control trials (RCTs). RCTs randomly divide patients into two (or more) groups and thus, based on statistical analyses, any difference in outcomes can be reasonably attributed to the difference in treatments. The outcomes thus measured are called endpoints. Selection of an appropriate and relevant endpoint is critical to design and conduct of a clinical trial.

In general, for a new treatment to be considered superior to an existing standard of care (SOC), RCTs are expected to show prolongation in overall survival (OS) and/or improvement in quality of life (QOL). Thus, OS and QOL are referred to as the clinical endpoints because they directly measure longer and better lives, respectively, which matter the most to both patients and physicians. However, some RCTs may measure outcomes other than these clinical endpoints. These alternative endpoints are called surrogate measures 1 , 2 or putative surrogate endpoints. 3 We are referring to them as “putative” surrogate endpoints because many outcomes currently labelled as surrogate endpoints don't meet the criteria for surrogacy.

An ideal surrogate endpoint should be easily measured, be a substitute for OS or QOL at an individual level, and between-group differences in these endpoints should also be able to predict between-group differences in OS or QOL reliably in a RCT. Moreover, the surrogate endpoint usually lies somewhere in the middle of the chain between the start of treatment and death and can be reached sooner than OS, potentially reducing trial completion time. 4 , 5 This is one of the primary reasons why oncology has exploded with trials using putative surrogate endpoints as primary endpoints in the past decade. 4 , 6 However, the use of surrogate measures for drug approval has its challenges as many of them do not reliably predict the patient-centric clinical outcomes, thus potentially leading to approval of ineffective drugs. 7

In this review, we will use the format of Frequently Asked Questions (FAQs) to define the common surrogate endpoints used in oncology, summarise the criteria proposed for their validation, highlight common issues in clinical trials using surrogate endpoints-both in advanced and adjuvant settings, and discuss possible solutions to advance the field to optimise trial design and ensure timely access to effective cancer medicines to patients.

FAQs on surrogate endpoints in oncology

How are various surrogate endpoints used in oncology trials defined?

Many surrogate endpoints have been used in oncology clinical trials and they can be classified in several ways- for example, surrogate endpoints in curative setting (neo-adjuvant and adjuvant), and those in advanced/metastatic settings ( Table 1 ). For advanced settings, they can be sub-divided into tumor response endpoints and time-to-event endpoints. Table 1 gives the general definitions of some common surrogate endpoints used in oncology.

In advanced setting, overall response rates (ORR)–also referred to as objective response rates in the literature- and progression-free survival (PFS) are the most commonly used endpoints, and their definitions are derived from the RECIST criteria. 8 RECIST defines complete response as the complete disappearance of all target lesions, partial response as shrinkage of the sum of maximum tumor diameters by at least 30%, and disease progression (PD) as increase in the sum of maximum tumor diameters by at least 20%, appearance of new lesions, or death. Change in tumor size that does not meet the criteria for PR or PD is classified as stable disease (SD). Percentage of patients who achieve a complete response is referred to as complete response rate (CR) and those that achieve a partial response is referred to as partial response rate (PR). ORR is the percentage of patients who have achieved complete or partial responses (ORR = CR plus PR), and duration of response (DOR) is the median time until which the response is sustained (i.e. the tumor does not start to grow again or metastasize after initially responding). PFS is the time from randomization until disease progression (PD) as defined above. One commonly reported, but in our opinion meaningless and even potentially misleading surrogate measure, is the clinical benefit rate (CBR) defined as the sum of ORR and SD rate, based on the false premise that anything short of PD is a therapeutic benefit.

There are usually no controversies in the definitions of these surrogate endpoints used in trials of advanced cancers. However, in curative settings, debates exist regarding the definitions of surrogate endpoints, sometimes even within the same tumor type, specially with regards to what constitutes an event ( Table 1 ). For example, whether second primary cancer (not directly linked to primary event) or in-situ recurrences (that can easily be treated and may not be linked to survival) should be included under disease free survival (DFS) events is debatable. Recurrence free survival (RFS) or invasive DFS (iDFS) are proposed alternative surrogate endpoints to DFS that differ only in regards to how these events are treated. 9 Some time to event endpoints like metastasis free survival (MFS) have been used specifically in prostate cancer and found to be a strong surrogate for OS. 10 Validation in other settings is desirable. Various initiatives like DATECAN and STEEP have attempted to homogenise these criteria within a given setting. 11 , 12 , 13 , 14

Tumor response endpoints in general reflect biological activity of the agent. They have been defined both radiologically and pathologically in the neoadjuvant setting (ORR; major pathological response, MPR; pathological complete response, pCR, measurable residual disease, MRD) and radiologically in the advanced setting (ORR, CBR) ( Table 1 ). Although these are commonly used as endpoints for early phase trials to establish preliminary activity, none of these endpoints have been validated as surrogates for OS or QOL at trial level and confirmatory evidence of clinical benefit should be required prior to approval by regulators and adoption in clinical practise.

How are surrogate endpoints validated?

If a putative surrogate measure is a valid surrogate endpoint, treatment effects on the surrogate endpoint is supposed to reliably predict treatment effects on OS. This is referred to as trial level validation of the endpoint. 15 This validation is usually conducted by determining the correlation coefficient between the treatment effects (differences between the arms in RCTs measured via median times or hazard ratios) on surrogate endpoint and the clinical endpoint across multiple trials in a meta-analysis (Figure). The German Institute for Quality and Efficiency in Health Care (IQWIG) has provided guidelines for defining the strength of association between the surrogate endpoint and the clinical endpoints. 16 Although these cut-offs are arbitrary, they present a useful metric for evaluating the validity of surrogate endpoints. Strength of association is considered strong with proven validity if lower limit of 95% confidence interval (CI) of the correlation coefficient is ≥0.85 and poor if upper limit of 95% CI is ≤0.7, anything in between is considered uncertain. This correlates to a coefficient of determination (R 2 ) of at least 0.7 for surrogacy to be proven, which in simple terms, means at least 70% of variation in OS is explained by the variation in surrogate endpoint. Putting these criteria into perspective, in a recent systematic review of surrogate and survival pairs, only 23% had strong correlation with OS 7 highlighting the “surrogate-survival” gap which currently exists in oncology clinical trials.

What is the difference between patient-level surrogate and trial-level surrogate?

A putative surrogate measure can be considered a validated surrogate endpoint for clinical trials when it shows utility as both a prognostic marker (patient-level surrogacy) and a predictive marker (trial-level surrogacy). 17 They should be prognostic in that they should be able to predict outcomes independent of treatment received. For example, pCR is a prognostic marker, which means irrespective of the neoadjuvant treatment received, achievement of pCR predicts for better patient outcomes, especially for HER-2+ and triple negative breast cancer. 18 , 19 , 20 This suggest that at a patient level, pCR is strongly corelated with long term outcomes like EFS/OS.

However, that alone does not make pCR a valid surrogate endpoint because surrogate validation requires validation not only as a patient-level surrogate but also as a trial-level surrogate. Predictive marker (or trial-level surrogate) means that the treatment effects on the surrogate should reliably predict treatment effects on the clinical endpoint. pCR unfortunately fails this criterion. That is, achieving pCR is better for a given patient irrespective of treatment received (good prognostic marker or patient-level surrogate) but treatments that improve pCR rates in trial do not necessarily improve EFS/OS (poor predictive marker or trial-level surrogate). 18 , 19 , 21 , 22 Similar findings of good patient level but poor trial-level surrogacy of pCR for OS was also observed in lung cancer. 23

ORR and CR reflect biological activity of agents and are prognostic for individual patient outcomes but poor trial level surrogates across multiple tumor sites. 24 , 25 , 26 , 27 , 28 They are reasonable endpoints for phase II studies to make go/no-go decisions but should not be typically used for drug approval or clinical decisions as they do not reliably predict OS.

What are the potential benefits of using surrogate endpoints?

The ability of surrogate endpoints like PFS and ORR to measure treatment efficacy earlier in the disease course than OS makes them attractive from a regulatory standpoint. This is considered important in oncology where treatments can take a long time to show their effect on clinical outcomes, and waiting for these outcomes can delay the approval of potentially life-saving treatments. However, previous research has shown that this reduction in study times is by an average of 11–19 months when ORR or PFS was used instead of OS as the primary endpoint. 1 This represents important time-savings but is not several years, as commonly believed. Surrogate endpoints may also improve efficiency of clinical trials as smaller sample sizes may be required to demonstrate clinically significant effects in surrogate endpoints than in OS or QOL, which may ultimately reduce trial cost and make them less resource intensive. 29

Furthermore, single-arm phase 1 and 2 trials necessarily need to employ surrogate endpoints for quick go/no-go decisions. Using surrogate endpoints is thus very reasonable in single arm trials. However, when approval decisions are made based on these trials, the lack of surrogacy becomes an important consideration. 30 Single arm trials require small sample size and thus, maybe relevant to answer questions in rare diseases. Thus, surrogates from single arm trials are used as justifications for drug approvals in rare cancers. However, studies have shown that most single-arm studies leading to FDA approval would have been feasible as RCTs within reasonable time frames. 31

Another purported benefit of using surrogate endpoints such as PFS is that it does not get diluted by post-protocol treatment effects. It is difficult to demonstrate an OS benefit for an experimental treatment in a disease with indolent biology, or one where survival post progression (SPP) is long and multiple treatment options are available after progression as all these factors tend to dilute the effect on OS. 32 , 33 , 34

From a regulatory perspective, surrogate endpoints can expedite the drug development process by allowing for shorter, smaller, and more cost-effective clinical trials. However, careful consideration on validation of these surrogates and conduct of appropriate confirmatory studies is essential to ensure that we are not misled by surrogates.

What strength of surrogacy is required for accelerated versus regular approval?

The US FDA has accepted surrogate endpoints for both accelerated and regular (or traditional) approval of cancer drugs. 35 For traditional approval, improvement in clinical endpoints is typically required, and therefore, if a surrogate endpoint is used for traditional approval, it needs to be validated for improving clinical endpoints. For the accelerated approval, the surrogate only needs to be reasonably likely to predict clinical outcomes, since the approval is conditional upon a confirmatory trial verifying the clinical benefit. Thus, in theory, the strength of validation of a surrogate should determine whether the drug receives traditional or accelerated approval.

However, studies 30 , 36 have shown that the FDA, in recent years, has offered regular or traditional approvals rather than offering accelerated approval based on surrogate endpoints. Indeed, from 2009 to 2018, although the proportion of cancer drugs approved based on surrogate endpoints increased, the proportion of accelerated approval decreased highlighting the fact that the distinction between surrogacy strength necessary for traditional versus accelerated approvals has vanished. 36 Often, regular approvals are granted based on only response rates from single-arm trials. 30 This is more common for oncology drugs than in other medical specialties. 37

Although the FDA has published a Table of Surrogate Endpoints that it will accept for approvals, distinction between endpoints suitable for regular vs accelerated approval is absent (except for pCR rates which is listed only under accelerated approval). 38 A study of cancer drugs that received accelerated approval and underwent confirmatory trials revealed that in 20% of cases, the confirmatory trial used the same surrogate endpoint as was used in the trial that led to accelerated approval highlighting lack of clarity and transparency regarding the surrogacy requirement for accelerated versus traditional approval. 39 A recent analysis of FDA approvals between 2001 and 2018 also highlighted that OS benefit was shown for only 32% of drugs receiving accelerated approval. Moreover, improved OS data was published much more promptly (median 1.5 years from initial results) than negative results (median 3.3 years) 40

Are we overusing and over-relying on surrogates for regulatory approvals?

Chen et al. analysed all cancer drug approvals granted by US FDA between 1992 and 2019 and found a significant increase in the use of surrogate measures between 2011 and 2019 compared to 2000–2010 for both regular (31.5% vs 18.6%) and accelerated approval (29.5 vs 13.3%). 6 For 28% of drugs approved, no OS or QOL data had been reported. Moreover, among the approvals with no or unknown corelation with OS, 53% were granted as regular approvals. Liu et al. reviewed the accelerated approval of cancer drugs between 2013 and 2023 and 98% of approvals were based on surrogate endpoints. 41 In 2018, 93% of approvals for adult solid tumors were based on surrogate endpoints. 36

Another problematic trend has been approvals based on response rates from single arm studies. 30 , 36 In addition to response rates being very poor predictors of clinical benefit, response rates and durations are often falsely inflated in single-arm trials versus their randomized counterparts. 42 , 43 In addition, concerns have been raised about relying on partial response as a marker of drug efficacy because even placebos can produce some responses in clinical trials. 44 , 45 However, CR from placebo is rare and maybe a more reliable marker of drug activity and needs to be further explored. 46

The granting of upfront regular (as opposed to accelerated) approval based on surrogate endpoints is referred to as premature regular approval. 47 Premature regular approvals are problematic because unlike accelerated approval, regular approvals do not have the safety net of a mandatory confirmatory trial to confirm clinical benefit. Thus, even if the drug was discovered not to improve any clinical outcomes in the future, it would be very difficult to get it withdrawn.

Irrespective of the type of approval pathway, accelerated or regular, studies have shown that 86% of drugs approved based on surrogate endpoints have not shown to improve overall survival several years post follow-up. 48 In addition, although our discussion of accelerated and regular approval concentrates on FDA's system, similar issues have been raised in other jurisdictions as well. 49 , 50 , 51 Granting early provisional approval based on surrogate endpoint and conforming clinical benefit on a clinical endpoint post-approval seems to provide the best balance for most systems.

Should we consider surrogate endpoints differently in the adjuvant setting?

Clinical trials in the adjuvant setting face unique challenges. First, any treatment in adjuvant setting is necessarily blind since response assessment is impossible in absence of visible tumor. Second, a majority of patients would have been cured without the need for adjuvant therapy and a percentage of patients would relapse despite the receipt of adjuvant therapy-the two groups who do not benefit but suffer harms from any adjuvant therapy. Third, any such potential utility of adjuvant therapy for the given patient can only be understood retrospectively at a distant time point in the future based on whether the cancer relapsed or not. Fourth, quality of life impact from adjuvant therapy can never be beneficial given the control arm would be not receiving this therapy and thus, not suffering the side effects of the therapy.

Therefore, the goal of any adjuvant therapy should necessarily be to improve survival and clinical decision making relying on surrogates alone would be very challenging. However, in the adjuvant setting, demonstrating an OS benefit takes a much longer time than in the metastatic setting making trials powered for OS longer and more expensive to conduct. Therefore, most recent trials use surrogate measures like DFS as primary endpoints. 52 Use of DFS in the adjuvant setting is justified in certain cancers where validation data exist, 53 , 54 , 55 , 56 , 57 however, the considerations for using surrogate endpoints in adjuvant studies require a much deeper discussion taking the principles of 3E's into consideration-evidence, ethics, and economics-as proposed by Gyawali and West. 58

Heterogeneity of the validity of surrogate endpoints-can we extrapolate from one setting to another?

Surrogacy validation is dynamic i.e. the strength of correlation for the same surrogate changes based not only on the tumor type but also between several subtypes of the same cancer, several lines of therapy or several types of therapy. For example, although DFS maybe a good surrogate endpoint in NSCLC for cytotoxic drugs, it failed to predict OS improvement in adjuvant immunotherapy trials. 55 , 59 In breast cancer, DFS is a validated surrogate for OS in HER2 positive subtype but not others. 53 However, this meta-analysis included patient data from older trials where event rate was higher compared to modern adjuvant studies. In recent adjuvant studies like APHINITY, no OS benefit has been shown at 8-year follow up despite early DFS benefit. 60 DFS has not been validated in TNBC or hormone receptor positive breast cancer, however, continues to be used in recent trials as the primary endpoint and as a basis for approval despite lack of OS benefit. 61 , 62 , 63 EFS is similar to DFS except that EFS is used in the neoadjuvant setting as patients are not disease free at that time, and also includes inability to undergo surgery as an event. In addition, although DFS is a validated endpoint for HER2 positive early breast cancer, the same validation does not hold true for EFS. Only moderate trial level associations were noted between EFS and OS for HER2 positive breast cancer in a meta-analysis by Gyawali et al. 64 Although individual level corelation were strong between 3-year EFS and OS for TNBC (R 2 = 0.82 (95% CI, 0.68–0.91) p < 0.01), trial level associations were moderate (R 2 of 0.64 (95% CI: 0.45–0.82), again highlighting the challenges in extrapolating data of surrogate endpoint validity across subtypes of the same tumor. 65

What is the strength of correlation for PFS, the most commonly used surrogate endpoint in trials of advanced cancer?

The validity of PFS as a surrogate for OS across different diseases is variable but is mostly weak except in some rare cases. 17 , 38 , 66 , 67 , 68 , 69 , 70 , 71 Strength of correlation is usually better in settings with very poor prognosis such as last line therapies of any cancer, but these are also the situations where OS events occur quickly rendering decisions based on PFS unhelpful.

Few studies have looked at the validity of PFS as a surrogate for OS in the era of targeted therapy and immunotherapy. A comprehensive review of literature by Hess et al. summarised the differential effects of biological/targeted agents on PFS and OS. 34 A striking finding was average benefit in both PFS and OS of 1.2 months across indications and trials, although this finding has numerous caveats. Among the 206 pairwise comparisons tested, in 39 (18.9%) there was an improvement in PFS but not in OS, whereas in 25 (12.1%), it was vice-versa. In 25 comparisons where the magnitude of PFS benefit was larger (>4 months), 9 did not show an OS benefit. Conversely, in a meta-analysis by Gyawali et al. evaluating trials of immunotherapy, OS benefit was observed in several trials despite lack of PFS benefit and PFS alone did not capture the effect in survival. 72 Zhang et al. did a meta-analysis of cancer immunotherapy trials and found that the intermediate endpoints ORR, disease control rate (DCR), and PFS have only a weak correlation with clinical outcomes. 73 The validity of surrogate endpoints in trials of PD1/PD-L1 inhibitors plus chemotherapy was evaluated by Villacampa et al. Thirty-nine RCTs studying then different immune checkpoint inhibitors was reviewed. PFS had a moderate association with OS with R 2 of 0.52. 74 Similar analysis by FDA have found weak to moderate corelation of ORR and PFS with OS in ICI trials and call for routine reporting of mature OS data in such studies. 75 The mechanism of immunotherapy drugs providing prolonged disease control in some patients despite early progression in others as well as the phenomena of pseudo-progression complicates the expected linear relationship between PFS and OS. Therefore, PFS may not be an adequate endpoint in immunotherapy trials because it may underestimate OS gains, unlike the case with cytotoxic agents where PFS overestimates OS gains. 76

There are also limitations among the validation studies. In an interesting analysis by Belin et al., 91 studies across cancer types evaluating PFS as a surrogate for OS were evaluated. They highlighted challenges in interpreting the results of surrogate validation studies as the analytical approaches used to determine coefficients of determination and corelation varied widely. However, even with conservative estimates and taking both individual and trial level corelation into account, in 32% of studies no quantitative argument of surrogacy was found even though authors claimed surrogacy of PFS for OS. Among 47 studies concluding PFS as a valid surrogate for OS, only 24 reported a R 2 > 0.6. 77

Why does PFS not corelate with OS?

There are several explanations for why PFS gains do not capture OS gains in cancer trials. First is the definition of PFS. Since progression is arbitrarily defined by RECIST criteria, there is no biological explanation why delaying tumor growth beyond an arbitrary 20% should correlate with OS. Second, any gains through delay in PFS from an earlier line treatment could be offset by differential survival post progression (SPP). A simulation analysis showed that corelation between PFS and OS was moderate when SPP was short (<12 months) but poor when SPP was long (>12 months). 32 However, when SPP is short, the time it takes to obtain OS results is short, thereby limiting the utility of surrogates. This is true for most aggressive cancers like hepatobiliary cancers, TNBC, small cell lung cancer and later line therapies of most other solid tumors.

Third, crossover from control to experimental arm at time of relapse can potentially dilute demonstration of survival gains from a given therapy. However, this finding would be useful by proving that the drug wouldn't be needed in earlier lines and could be reserved for post progression among only those who need it. For example, SONIA trial showed that although PFS would improve by adding Palbociclib to endocrine therapy as first-line treatment for hormone positive breast cancer, similar OS could be achieved by offering Palbociclib second line to only those who progressed, thereby offering lesser toxicity, better QOL, and cheaper treatment costs. 78 Indeed, crossover should be mandated when doing a trial of a drug in an earlier line. 79

Fourth, differential or informative or unequal censoring can also confound the relationship between PFS and OS. This usually happens when the experimental treatment is toxic leading to a large imbalance in discontinuation rates, higher dropouts in experimental arm, and nullifies the effect of randomisation as the trial ends up comparing patients who were able to tolerate the more toxic medication to the standard of care group. 80 , 81 If sicker patients in the experimental arm drop out (due to lack of efficacy and/or toxicity) and healthier patients in the control arm drop out (due to disappointment with being randomized to control treatment and seeking experimental drugs outside, who are also better in terms of socio-economic status to be able to pursue this), then effectively randomization becomes void since the trial would end up comparing healthiest of experimental arm patients with sickest of control arm patients, instead of comparing random cohorts of patients. A simulation study shows that a toxic but not-effective drug could falsely show positive PFS gains based on informative censoring alone. 82 An interesting analysis by Gilboa et al. reported significant censoring imbalance in 15% of 55 trials that reported discordant results between a surrogate measure and OS; among these 50% lost statistical significance on reverse K-M analysis that accounted for informative censoring. 83 A classic example of this phenomenon is seen in the BOLERO-2 study which compared everolimus plus exemestane to exemestane alone in 2nd/3rd line setting in patients with metastatic hormone receptor positive breast cancer. Although a significant PFS benefit was observed with combination, there was no OS benefit which was likely explained by significantly higher rates of discontinuation in everolimus arm (29% vs 5%) leading to informative censoring. 80 , 84 , 85

Assessment of PFS is also subject to a number of biases, especially in open label studies. 86 Other errors and biases in assessment of PFS, and why relying on PFS for cancer drugs registration could be misleading, have been outlined in a previous essay. 87

Evaluation-time bias occurs when progression is evaluated more or less frequently in one arm compared to the other depending on treatment assignment, thus influencing the PFS estimates. This could arise when an investigator considers an experimental treatment to be better than control/placebo and there may be a tendency to evaluate patients in the control arm at shorter intervals in order to declare progression and get them on the experimental arm (if cross over is allowed) or subsequent therapy (if not allowed). This is especially problematic given the frequency of trials with sub-standard control arms. 88 , 89 For example, the 3 days benefit in median PFS in NALA trial could be explained away with evaluation-time bias alone. 90 PFS also suffers from reader-evaluation bias as determination of progression on scans is often subjective and this issue is often exaggerated in unblinded studies. 91 When there is discrepancy between local investigator and blinded independent central review, the true median PFS may lie somewhere between the two. 92 OS is less affected by these limitations.

Does PFS correlate with quality of life (QOL)?

Unfortunately, PFS has not been shown to corelate with QOL either, the second clinical endpoint that is most meaningful for patients. Hwang and Gyawali examined 352 cancer RCTs and found that only 42% reported QOL data, and that the correlation between PFS gains and improved QOL was weak (r = 0.34, AUC = 0.72), both for global QOL or any QOL domain. 93 The findings were consistent in another study of cancer trials. 94 Another study found that among 45 trials that demonstrated improved PFS, 15% in fact worsened QOL while in no case a drug that improved OS worsened QOL. 95

Do patients and public understand PFS correctly?

There remains a lot of misunderstanding among the patients and the public about what PFS actually means. In an experiment of direct-to-consumer advertising of cancer drugs, participants could not differentiate between ORR, PFS and OS endpoints regarding whether the drug helped patients live longer. Although participants understood the differences between ORR, PFS, and OS, they still believed that patients lived longer when ORR and specially, PFS was improved. Even after adding a disclosure statement that specifically mentioned that the effects on OS were unknown, a substantial percentage of participants continued to believe that a drug improved ORR or PFS also improved survival. 96 To help improve clarity around the unknown effects on OS based on PFS findings alone, renaming PFS to progression-free interval (PFI) has been proposed, since the “survival” in PFS may confuse patients, public, and physicians that improving PFS automatically meant improved OS. 97

A recent study on patient preferences also highlights that when properly understood, >50% patients do not prefer drugs that improve only PFS (irrespective of magnitude) but do not improve either QOL or OS. 98

Why do regulators continue to rely on PFS?

In some cases, there are reasons to believe that regulators look at PFS not as a surrogate but as a clinical endpoint in itself and provide ‘premature regular approval’ as discussed before. However, if true, this is misguided. On several occasions, FDA has offered regular approval based on PFS – instead of accelerated approval-even when the OS results were negative (neratinib in breast cancer) or would have been available within a year (e.g. everolimus or margetuximab in breast cancer). 84 , 85 , 99 , 100 , 101 Therefore, based on all available evidence, including lack of surrogacy, patient preferences, and availability of OS data within a reasonable timeframe, considering PFS a clinical endpoint would be deeply flawed and misguided.

What is the way forward

Various issues with surrogate endpoints have been highlighted in this review ( Fig. 1 ). It is clear that prior to using surrogate measures as endpoints for randomised trials or as basis of approval of cancer drugs, they need to be validated appropriately. The strength of surrogacy required for accelerated versus traditional approval should be transparent. A prognostic marker (patient-level surrogate) should not be mistaken for a trial-level surrogate endpoint.

We highlighted several pitfalls with using surrogate endpoints as well as the complexities with surrogacy validation studies. Indeed, validating a surrogate endpoint is not an easy task. And despite validation, a new trial may show discordance between surrogate and survival outcomes. Given that surrogacy varies based on several parameters-line of therapy, disease subtype, nature of treatment, etc-there is a risk that any validation study is already outdated by the time it is published. Indeed, in an analysis of all bevacizumab trials in metastatic breast cancer, the conclusion was that the validation of PFS as a surrogate for OS was uncertain despite 7 RCTs available. 102 If the surrogacy cannot be conclusively inferred even after 7 RCTs, we believe it is far more efficient for the system to conduct one proper RCT powered for OS than to keep guessing about the strength of surrogacy after multiple RCTs.

In fact, a new guidance from the FDA regarding accelerated approval suggests a ‘one trial approach’ with the trial powered for both a surrogate endpoint (like ORR or PFS) and clinical benefit endpoint like OS with adaptive design allowing for dynamic sample size assessments and control of type 1 error rates. 103 Such a trial will not only provide information on OS but will also simultaneously prove/disprove the surrogacy. The sample size to run such a trial will also be less than the need to conduct 2 different trials for accelerated and regular approvals under the current approach. This could speed up efficiency and address the current problem with surrogacy simultaneously. However, it is unclear whether the FDA will mandate that OS be used as the endpoint for the confirmatory phase of this one-trial design. 104

Logistically, it is efficient to conduct a proper RCT with OS endpoint in the last-line therapy setting of a clinical trial given that survival endpoints are achieved early. Also, as there is no standard therapy, clinical outcomes would not be affected by inadequate post-protocol therapy in the control arm. Thus, both crossover and long SPP issues wouldn't apply. Also, because PFS cannot predict QOL outcomes, QOL should be measured separately rather than predicting QOL outcomes based on PFS.

Certain cases deserve special attention. In clinical trials studying combination of drugs that would usually be used in sequence, or trials studying maintenance therapies where a drug is given for a prolonged period of time where the SOC would normally be observation alone, PFS endpoint is never sufficient. PFS endpoint alone is not sufficient as sequential usage may lead to similar total PFS as upfront use of all drugs. A + B combination having longer PFS than single agent of A or B is expected but an OS benefit is needed to justify the added therapeutic and financial burden of additional therapy on all patients compared to using these agents sequentially among those who progressed. 105 Similarly, PFS gains with a maintenance therapy versus no treatment is also quite expected, but this added therapeutic burden can be justified only if patients live longer as a result.

We also believe that, given how PFS is arbitrarily defined based on RECIST criteria, symptomatic PFS maybe more meaningful than mere imaging based PFS. Validation studies incorporating symptomatic PFS into clinical trials are needed to analyse if symptomatic PFS performs better than PFS as a surrogate endpoint. In addition, any trial with PFS endpoint must mandatorily be blinded.

With regards to responses, we can safely conclude that ORR should seldom be relied upon as a surrogate endpoint. However, complete response rates could potentially be used as a surrogate for accelerated approval given complete responses (but not partial responses) are almost always representative of the beneficial effect of the drugs.

Even when these surrogates are used, they should only lead to accelerated approval. Accelerated approval ensures early access to the drug, while maintaining the safety net to withdraw the drug if found to be ineffective in confirmatory trials.

If the endpoint is OS or validated surrogate, can we be confident in the results?

When evaluating trials, surrogacy is only one part of critical appraisal. Multiple other points need to be examined including appropriateness of control arm, presence of informative censoring in the experimental arm, proportion of patients receiving the active drug at relapse and finally consistency with other similar trials. These points have been highlighted by the position paper of ESMO-MCBS as well, 106 since appraisal of magnitude of clinical benefit entails consideration of factors including but not limited to surrogacy alone. Moreover, overpowering trials by accruing large number of patients to measure small differences which may be statistically significant but not clinically meaningful exposes a large number of patients to potentially toxic treatments for no clinical benefit and may lead to net harm.

Conclusions

While surrogate endpoints help trials to complete early, they can also be potentially misleading. Although surrogate measures have a role in offering early access to therapies via accelerated approval, putative surrogate endpoints should be properly validated before accepting them as endpoints for traditional approvals. A more efficient approach to drug development would be to conduct RCTs powered for OS that definitively answers the effect of the drug on clinical endpoints rather than to conduct multiple trials with surrogate endpoints and resulting uncertainty regarding the effect of the drug on survival. Alternatively, as proposed in the new FDA guidance, a one-trial approach of conducting an RCT that is powered for OS from the outset but the same RCT sufficing for early accelerated approval based on responses/PFS and mature regular approval based on OS offers a reasonable balance between speed and efficacy. OS should also be the defacto endpoint specially in later line settings or where the median OS is less than a year. PFS should be better renamed as PFI given the public and patient confusion regarding the phrase. QOL outcomes should be directly tested in trials and not inferred based on surrogates.

Contributors

BG conceptualized the study. AM and MSK wrote the first draft. SD and KW contributed to literature search. BG supervised and mentored the project. All authors contributed to manuscript writing, editing, and revising and agreed on the final submission.

Declaration of interests

BG has received consulting fees from Vivio Health, and holds stock options in OneCell Diagnostics, both unrelated to this manuscript. All other authors have no interests to declare.

Acknowledgements

Dr. Gyawali gratefully acknowledges salary support from Ontario Institute for Cancer Research funded by the Government of Ontario. The opinions expressed in this article are those of the authors and do not necessarily represent that of the government.

This work has not received any specific funding.

References

1. Chen E.Y., Joshi S.K., Tran A., Prasad V. Estimation of study time reduction using surrogate end points rather than overall survival in oncology clinical trials. JAMA Intern Med. 2019;179(5):642–647. doi: 10.1001/jamainternmed.2018.8351.

2. Patil S., Agarwal V., Drupad H.S. Significance of emerging clinical oncology endpoints in support of overall survival. Indian J Cancer. 2022;59(Supplement):S106–S118. doi: 10.4103/ijc.IJC_1417_20.

3. Del Paggio J.C., Berry J.S., Hopman W.M., et al. Evolution of the randomized clinical trial in the era of precision oncology. JAMA Oncol. 2021;7(5):728–734. doi: 10.1001/jamaoncol.2021.0379.

4. Kemp R., Prasad V. Surrogate endpoints in oncology: when are they acceptable for regulatory and clinical decisions, and are they currently overused? BMC Med. 2017;15:134. doi: 10.1186/s12916-017-0902-9.

5. Wright K., Mittal A., Gyawali B. Surrogate endpoints for HTA decisions of breast cancer drugs: utility and pitfalls. Curr Opin Oncol. 2023;35(6):513–521. doi: 10.1097/CCO.0000000000000984.

6. Chen E.Y., Haslam A., Prasad V. FDA acceptance of surrogate end points for cancer drug approval: 1992-2019. JAMA Intern Med. 2020;180(6):912–914. doi: 10.1001/jamainternmed.2020.1097.

7. Prasad V., Kim C., Burotto M., Vandross A. The strength of association between surrogate end points and survival in oncology: a systematic review of trial-level meta-analyses. JAMA Intern Med. 2015;175(8):1389–1398. doi: 10.1001/jamainternmed.2015.2829.

8. Eisenhauer E.A., Therasse P., Bogaerts J., et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1) Eur J Cancer. 2009;45(2):228–247. doi: 10.1016/j.ejca.2008.10.026.

9. Chua Y.J., Sargent D., Cunningham D. Definition of disease-free survival: this is my truth–show me yours. Ann Oncol. 2005;16(11):1719–1721. doi: 10.1093/annonc/mdi373.

10. Xie W., Regan M.M., Buyse M., et al. Metastasis-free survival is a strong surrogate of overall survival in localized prostate cancer. J Clin Orthod. 2017;35(27):3097–3104. doi: 10.1200/JCO.2017.73.9987.

11. Gourgou-Bourgade S., Cameron D., Poortmans P., et al. Guidelines for time-to-event end point definitions in breast cancer trials: results of the DATECAN initiative (Definition for the Assessment of Time-to-event Endpoints in CANcer trials) Ann Oncol. 2015;26(5):873–879. doi: 10.1093/annonc/mdv106.

12. Bellera C.A., Pulido M., Gourgou S., et al. Protocol of the Definition for the Assessment of Time-to-event Endpoints in CANcer trials (DATECAN) project: formal consensus method for the development of guidelines for standardised time-to-event endpoints' definitions in cancer clinical trials. Eur J Cancer. 2013;49(4):769–781. doi: 10.1016/j.ejca.2012.09.035.

13. Cohen R., Vernerey D., Bellera C., et al. Guidelines for time-to-event end-point definitions in adjuvant randomised trials for patients with localised colon cancer: results of the DATECAN initiative. Eur J Cancer. 2020;130:63–71. doi: 10.1016/j.ejca.2020.02.009.

14. Tolaney S.M., Garrett-Mayer E., White J., et al. Updated standardized definitions for efficacy end points (STEEP) in adjuvant breast cancer clinical trials: STEEP version 2.0. J Clin Orthod. 2021;39(24):2720–2731. doi: 10.1200/JCO.20.03613.

15. Buyse M., Molenberghs G., Burzykowski T., Renard D., Geys H. The validation of surrogate endpoints in meta-analyses of randomized experiments. Biostatistics. 2000;1(1):49–67. doi: 10.1093/biostatistics/1.1.49.

16. Institute for quality and efficiency in Health care: executive summaries Institute for Quality and Efficiency in Health Care (IQWiG) 2011. Validity of surrogate endpoints in oncology. https://www.ncbi.nlm.nih.gov/books/NBK198799/ [cited 2023 Apr 28]. Available from:

17. Buyse M., Sargent D.J., Grothey A., Matheson A., de Gramont A. Biomarkers and surrogate end points--the challenge of statistical validation. Nat Rev Clin Oncol. 2010;7(6):309–317. doi: 10.1038/nrclinonc.2010.43.

18. Cortazar P., Zhang L., Untch M., et al. Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. Lancet. 2014;384(9938):164–172. doi: 10.1016/S0140-6736(13)62422-8.

19. Broglio K.R., Quintana M., Foster M., et al. Association of pathologic complete response to neoadjuvant therapy in HER2-positive breast cancer with long-term outcomes: a meta-analysis. JAMA Oncol. 2016;2(6):751–760. doi: 10.1001/jamaoncol.2015.6113.

20. von Minckwitz G., Untch M., Blohmer J.U., et al. Definition and impact of pathologic complete response on prognosis after neoadjuvant chemotherapy in various intrinsic breast cancer subtypes. J Clin Oncol. 2012;30(15):1796–1804. doi: 10.1200/JCO.2011.38.8595.

21. Berruti A., Amoroso V., Gallo F., et al. Pathologic complete response as a potential surrogate for the clinical outcome in patients with breast cancer after neoadjuvant therapy: a meta-regression of 29 randomized prospective studies. J Clin Oncol. 2014;32(34):3883–3891. doi: 10.1200/JCO.2014.55.2836.

22. Squifflet P., Saad E.D., Loibl S., et al. Re-evaluation of pathologic complete response as a surrogate for event-free and overall survival in human epidermal growth factor receptor 2-positive, early breast cancer treated with neoadjuvant therapy including anti-human epidermal growth factor receptor 2 therapy. J Clin Oncol. 2023;28 doi: 10.1200/JCO.22.02363.

23. Rosner S., Liu C., Forde P.M., Hu C. Association of pathologic complete response and long-term survival outcomes among patients treated with neoadjuvant chemotherapy or chemoradiotherapy for NSCLC: a meta-analysis. JTO Clin Res Rep. 2022;3(9) doi: 10.1016/j.jtocrr.2022.100384.

24. Kok P.S., Yoon W.H., Lord S., Marschner I., Friedlander M., Lee C.K. Tumor response end points as surrogates for overall survival in immune checkpoint inhibitor trials: a systematic review and meta-analysis. JCO Precis Oncol. 2021;5:1151–1159. doi: 10.1200/PO.21.00108.

25. Navani V., Ernst M., Wells J.C., et al. Imaging response to contemporary immuno-oncology combination therapies in patients with metastatic renal cell carcinoma. JAMA Netw Open. 2022;5(6) doi: 10.1001/jamanetworkopen.2022.16379.

26. Hua T., Gao Y., Zhang R., Wei Y., Chen F. Validating ORR and PFS as surrogate endpoints in phase II and III clinical trials for NSCLC patients: difference exists in the strength of surrogacy in various trial settings. BMC Cancer. 2022;22(1):1022. doi: 10.1186/s12885-022-10046-z.

27. Cooper K., Tappenden P., Cantrell A., Ennis K. A systematic review of meta-analyses assessing the validity of tumour response endpoints as surrogates for progression-free or overall survival in cancer. Br J Cancer. 2020;123(11):1686–1696. doi: 10.1038/s41416-020-01050-w.

28. Haslam A., Hey S.P., Gill J., Prasad V. A systematic review of trial-level meta-analyses measuring the strength of association between surrogate end-points and overall survival in oncology. Eur J Cancer. 2019;106:196–211. doi: 10.1016/j.ejca.2018.11.012.

29. Ciani O., Manyara A.M., Chan A.W., Taylor R.S. On behalf of the SPIRIT-SURROGATE/CONSORT-SURROGATE project group. Surrogate endpoints in trials: a call for better reporting. Trials. 2022;23(1):991. doi: 10.1186/s13063-022-06904-7.

30. Agrawal S., Arora S., Amiri-Kordestani L., et al. Use of single-arm trials for US food and drug administration drug approval in oncology, 2002-2021. JAMA Oncol. 2023;9(2):266–272. doi: 10.1001/jamaoncol.2022.5985.

31. Rittberg R., Czaykowski P., Niraula S. Feasibility of randomized controlled trials for cancer drugs approved by the food and drug administration based on single-arm studies. JNCI Cancer Spectr. 2021;5(4) doi: 10.1093/jncics/pkab061.

32. Amir E., Seruga B., Kwong R., Tannock I.F., Ocaña A. Poor correlation between progression-free and overall survival in modern clinical trials: are composite endpoints the answer? Eur J Cancer. 2012;48(3):385–388. doi: 10.1016/j.ejca.2011.10.028.

33. Matulonis U.A., Oza A.M., Ho T.W., Ledermann J.A. Intermediate clinical endpoints: a bridge between progression-free survival and overall survival in ovarian cancer trials. Cancer. 2015;121(11):1737–1746. doi: 10.1002/cncr.29082.

34. Hess L.M., Brnabic A., Mason O., Lee P., Barker S. Relationship between progression-free survival and overall survival in randomized clinical trials of targeted and biologic agents in oncology. J Cancer. 2019;10(16):3717–3727. doi: 10.7150/jca.32205.

35. Clinical trial endpoints for the approval of cancer drugs and biologics. 2018.

36. Gyawali B., Sharma S., Booth C.M. Is the number of cancer drug approvals a surrogate for regulatory success? J Cancer Policy. 2019;22

37. Mitra-Majumdar M., Gunter S.J., Kesselheim A.S., et al. Analysis of supportive evidence for US food and drug administration approvals of novel drugs in 2020. JAMA Netw Open. 2022;5(5) doi: 10.1001/jamanetworkopen.2022.12454.

38. Gyawali B., Hey S.P., Kesselheim A.S. Evaluating the evidence behind the surrogate measures included in the FDA's table of surrogate endpoints as supporting approval of cancer drugs. eClinicalMedicine. 2020;21 doi: 10.1016/j.eclinm.2020.100332.

39. Gyawali B., Hey S.P., Kesselheim A.S. Assessment of the clinical benefit of cancer drugs receiving accelerated approval. JAMA Intern Med. 2019;179(7):906–913. doi: 10.1001/jamainternmed.2019.0462.

40. Naci H., Zhang Y., Woloshin S., Guan X., Xu Z., Wagner A.K. Overall survival benefits of cancer drugs initially approved by the US Food and Drug Administration on the basis of immature survival data: a retrospective analysis. Lancet Oncol. 2024;25(6):760–769. doi: 10.1016/S1470-2045(24)00152-9. https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(24)00152-9/fulltext [cited 2024 May 23]; Available from:

41. Liu I.T.T., Kesselheim A.S., Cliff E.R.S. Clinical benefit and regulatory outcomes of cancer drugs receiving accelerated approval. JAMA. 2024;331(17):1471–1479. doi: 10.1001/jama.2024.2396.

42. Gyawali B., D’Andrea E., Franklin J.M., Kesselheim A.S. Response rates and durations of response for biomarker-based cancer drugs in nonrandomized versus randomized trials. J Natl Compr Canc Netw. 2020;18(1):36–43. doi: 10.6004/jnccn.2019.7345.

43. Zia M.I., Siu L.L., Pond G.R., Chen E.X. Comparison of outcomes of phase II studies and subsequent randomized control studies using identical chemotherapeutic regimens. J Clin Oncol. 2005;23(28):6982–6991. doi: 10.1200/JCO.2005.06.679.

44. Gleave M.E., Elhilali M., Fradet Y., et al. Interferon gamma-1b compared with placebo in metastatic renal-cell carcinoma. N Engl J Med. 1998;338(18):1265–1271. doi: 10.1056/NEJM199804303381804.

45. Gounder M.M., Mahoney M.R., Van Tine B.A., et al. Sorafenib for advanced and refractory desmoid tumors. N Engl J Med. 2018;379(25):2417–2428. doi: 10.1056/NEJMoa1805052.

46. Sachdev A., Sharpe I., Bowman M., Booth C.M., Gyawali B. Objective response rate of placebo in randomized controlled trials of anticancer medicines. eClinicalMedicine. 2023;55 doi: 10.1016/j.eclinm.2022.101753.

47. Medscape Why is the FDA granting regular approvals prematurely? https://www.medscape.com/viewarticle/983961 [cited 2023 Jul 25] Available from:

48. Elbaz J., Haslam A., Prasad V. An empirical analysis of overall survival in drug approvals by the US FDA (2006-2023) Cancer Med. 2024;13(8) doi: 10.1002/cam4.7190.

49. Maeda H., Shingai R., Takeda K., Hara A., Murai Y., Ofuchi M. Assessment of surrogate end point trends in clinical trials to approve oncology drugs from 2001 to 2020 in Japan. JAMA Netw Open. 2023;6(4) doi: 10.1001/jamanetworkopen.2023.8875.

50. Avorn J. Surrogate measures of drug efficacy—a finger pointing at the moon. JAMA Netw Open. 2023;6(4) doi: 10.1001/jamanetworkopen.2023.8835.

51. Pavic M., Wolfer A., Li Q., et al. Regulatory challenges: is a surrogate end point instead of overall survival enough for regulatory approval of (Neo)Adjuvant cancer treatment? The swissmedic perspective. J Clin Orthod. 2023;41(32):4973–4975. doi: 10.1200/JCO.23.01442.

52. Robinson A.G., Booth C.M., Eisenhauer E.A. Disease-free survival as an end-point in the treatment of solid tumours – perspectives from clinical trials and clinical practice. Eur J Cancer. 2014;50(13):2298–2302. doi: 10.1016/j.ejca.2014.05.016.

53. Saad E.D., Squifflet P., Burzykowski T., et al. Disease-free survival as a surrogate for overall survival in patients with HER2-positive, early breast cancer in trials of adjuvant trastuzumab for up to 1 year: a systematic review and meta-analysis. Lancet Oncol. 2019;20(3):361–370. doi: 10.1016/S1470-2045(18)30750-2.

54. Yin J., Salem M.E., Dixon J.G., et al. Reevaluating disease-free survival as an endpoint vs overall survival in stage III adjuvant colon cancer trials. J Natl Cancer Inst. 2022;114(1):60–67. doi: 10.1093/jnci/djab187.

55. Mauguen A., Pignon J.P., Burdett S., et al. Surrogate endpoints for overall survival in chemotherapy and radiotherapy trials in operable and locally advanced lung cancer: a re-analysis of meta-analyses of individual patients' data. Lancet Oncol. 2013;14(7):619–626. doi: 10.1016/S1470-2045(13)70158-X.

56. Nie R.C., Zou X.B., Yuan S.Q., et al. Disease-free survival as a surrogate endpoint for overall survival in adjuvant trials of pancreatic cancer: a meta-analysis of 20 randomized controlled trials. BMC Cancer. 2020;20:421. doi: 10.1186/s12885-020-06910-5.

57. Untch M., Pérol D., Mayer E.L., et al. Disease-free survival as a surrogate for overall survival in HR+/HER2– early breast cancer: a correlation analysis. Eur J Cancer. 2024;202 doi: 10.1016/j.ejca.2024.113977.

58. Gyawali B., West H. Lessons from ADAURA on adjuvant cancer drug trials: evidence, ethics, and economics. J Clin Oncol. 2020;39(3):175–177. doi: 10.1200/JCO.20.01762. https://ascopubs.org/doi/pdf/10.1200/JCO.20.01762?role=tab [cited 2023 May 21]; Available from:

59. Li Y., Yuan S., Chen Y., et al. Evaluating relapse-free survival as an endpoint for overall survival in adjuvant immunotherapy trials. J Natl Cancer Inst. 2023;30:djad125. doi: 10.1093/jnci/djad125.

60. Piccart M., Procter M., Fumagalli D., et al. Adjuvant pertuzumab and trastuzumab in early HER2-positive breast cancer in the APHINITY trial: 6 Years' follow-up. J Clin Oncol. 2021;39(13):1448–1457. doi: 10.1200/JCO.20.01204. https://ascopubs.org/doi/pdf/10.1200/JCO.20.01204 [cited 2022 Feb 24]; Available from:

61. Johnston S.R.D., Harbeck N., Hegg R., et al. Abemaciclib combined with endocrine therapy for the adjuvant treatment of HR+, HER2-, node-positive, high-risk, early breast cancer (monarchE) J Clin Oncol. 2020;38(34):3987–3998. doi: 10.1200/JCO.20.02514.

62. Johnston S.R.D., Toi M., O'Shaughnessy J., et al. Abemaciclib plus endocrine therapy for hormone receptor-positive, HER2-negative, node-positive, high-risk early breast cancer (monarchE): results from a preplanned interim analysis of a randomised, open-label, phase 3 trial. Lancet Oncol. 2023;24(1):77–90. doi: 10.1016/S1470-2045(22)00694-5.

63. Research C for DE . FDA; 2023. FDA D.I.S.C.O. Burst Edition: FDA approval of Verzenio (abemaciclib) with endocrine therapy for patients with HR-positive, HER2-negative, node-positive, early breast cancer. https://www.fda.gov/drugs/resources-information-approved-drugs/fda-disco-burst-edition-fda-approval-verzenio-abemaciclib-endocrine-therapy-patients-hr-positive [cited 2023 May 11]; Available from:

64. Gyawali B., D'Andrea E., Franklin J.M., Kesselheim A.S. A correlation analysis to assess event-free survival as a trial-level surrogate for overall survival in early breast cancer. eClinicalMedicine. 2021;32:100730. doi: 10.1016/j.eclinm.2021.100730. https://www.thelancet.com/journals/eclinm/article/PIIS2589-5370(21)00010-9/fulltext [cited 2023 May 12]; Available from:

65. Fasching P.A., Huang M., Haiderali A., et al. 186P Evaluation of event-free survival as a surrogate for overall survival in early-stage triple-negative breast cancer following neoadjuvant therapy. Ann Oncol. 2022;33

66. Hirai T., Nemoto A., Ito Y., Matsuura M. Meta-analyses on progression-free survival as a surrogate endpoint for overall survival in triple-negative breast cancer. Breast Cancer Res Treat. 2020;181(1):189–198. doi: 10.1007/s10549-020-05615-4.

67. Courtinard C., Gourgou S., Jacot W., et al. Association between progression-free survival and overall survival in women receiving first-line treatment for metastatic breast cancer: evidence from the ESME real-world database. BMC Med. 2023;21(1):87. doi: 10.1186/s12916-023-02754-5.

68. Buyse M., Burzykowski T., Carroll K., et al. Progression-free survival is a surrogate for survival in advanced colorectal cancer. J Clin Oncol. 2007;25(33):5218–5224. doi: 10.1200/JCO.2007.11.8836.

69. Shameer K., Zhang Y., Jackson D., et al. Correlation between early endpoints and overall survival in non-small-cell lung cancer: a trial-level meta-analysis. Front Oncol. 2021;11 doi: 10.3389/fonc.2021.672916.

70. Blumenthal G.M., Karuri S.W., Zhang H., et al. Overall response rate, progression-free survival, and overall survival with targeted and standard therapies in advanced non-small-cell lung cancer: US Food and Drug Administration trial-level and patient-level analyses. J Clin Oncol. 2015;33(9):1008–1014. doi: 10.1200/JCO.2014.59.0489.

71. Paoletti X., Lewsley L.A., Daniele G., et al. Assessment of progression-free survival as a surrogate end point of overall survival in first-line treatment of ovarian cancer: a systematic review and meta-analysis. JAMA Netw Open. 2020;3(1) doi: 10.1001/jamanetworkopen.2019.18939.

72. Gyawali B., Hey S.P., Kesselheim A.S. A comparison of response patterns for progression-free survival and overall survival following treatment for cancer with PD-1 inhibitors: a meta-analysis of correlation and differences in effect sizes. JAMA Netw Open. 2018;1(2) doi: 10.1001/jamanetworkopen.2018.0416.

73. Zhang Z., Pan Q., Lu M., Zhao B. Intermediate endpoints as surrogates for outcomes in cancer immunotherapy: a systematic review and meta-analysis of phase 3 trials. eClinicalMedicine. 2023;63 doi: 10.1016/j.eclinm.2023.102156.

74. G V., P C.M., V N., C V., R D. Comprehensive evaluation of surrogate endpoints to predict overall survival in trials with PD1/PD-L1 immune checkpoint inhibitors plus chemotherapy. Cancer Treat Rev. 2023;116:102542. doi: 10.1016/j.ctrv.2023.102542. https://pubmed.ncbi.nlm.nih.gov/37003083/ [cited 2024 May 16]. Available from:

75. Goulart B.H.L., Mushti S.L., Chatterjee S., et al. Correlations of response rate and progression-free survival with overall survival in immunotherapy trials for metastatic non-small-cell lung cancer: an FDA pooled analysis. Lancet Oncol. 2024;25(4):455–462. doi: 10.1016/S1470-2045(24)00040-8.

76. Chan E., Quinn C., Hirji I., et al. Alternative metrics for assessing clinical benefit with immunotherapy in oncology. Oncoimmunology. 2018;8(10) doi: 10.1080/2162402X.2017.1343774.

77. Belin L., Tan A., De Rycke Y., Dechartres A. Progression-free survival as a surrogate for overall survival in oncology trials: a methodological systematic review. Br J Cancer. 2020;122(11):1707–1714. doi: 10.1038/s41416-020-0805-y.

78. Primary outcome analysis of the phase 3 SONIA trial (BOOG 2017-03) on selecting the optimal position of cyclin-dependent kinases 4 and 6 (CDK4/6) inhibitors for patients with hormone receptor-positive (HR+), HER2-negative (HER2-) advanced breast cancer (ABC) J Clin Oncol. 2023 https://ascopubs.org/doi/10.1200/JCO.2023.41.17_suppl.LBA1000 [cited 2024 May 23]. Available from:

79. Gyawali B. Problematic crossovers in cancer drug trials. Nat Rev Clin Oncol. 2023;20(12):815–816. doi: 10.1038/s41571-023-00805-7.

80. Templeton A.J., Ace O., Amir E., et al. Influence of censoring on conclusions of trials for women with metastatic breast cancer. Eur J Cancer. 2015;51(6):721–724. doi: 10.1016/j.ejca.2014.12.016.

81. Templeton A.J., Amir E., Tannock I.F. Informative censoring — a neglected cause of bias in oncology trials. Nat Rev Clin Oncol. 2020;17(6):327–328. doi: 10.1038/s41571-020-0368-0.

82. Tannock I.F., Pond G.R., Booth C.M. Biased evaluation in cancer drug trials-how use of progression-free survival as the primary end point can mislead. JAMA Oncol. 2022;8(5):679–680. doi: 10.1001/jamaoncol.2021.8206.

83. Gilboa S., Pras Y., Mataraso A., Bomze D., Markel G., Meirson T. Informative censoring of surrogate end-point data in phase 3 oncology trials. Eur J Cancer. 2021;153:190–202. doi: 10.1016/j.ejca.2021.04.044.

84. Piccart M., Hortobagyi G.N., Campone M., et al. Everolimus plus exemestane for hormone-receptor-positive, human epidermal growth factor receptor-2-negative advanced breast cancer: overall survival results from BOLERO-2†. Ann Oncol. 2014;25(12):2357–2362. doi: 10.1093/annonc/mdu456.

85. Baselga J., Campone M., Piccart M., et al. Everolimus in postmenopausal hormone-receptor–positive advanced breast cancer. N Engl J Med. 2012;366(6):520–529. doi: 10.1056/NEJMoa1109653.

86. Freidlin B., Korn E.L., Hunsberger S., Gray R., Saxman S., Zujewski J.A. Proposal for the use of progression-free survival in unblinded randomized trials. J Clin Oncol. 2007;25(15):2122–2126. doi: 10.1200/JCO.2006.09.6198.

87. Booth C.M., Eisenhauer E.A., Gyawali B., Tannock I.F. Progression-free survival should not be used as a primary end point for registration of anticancer drugs. J Clin Oncol. 2023;41(32):4968–4972. doi: 10.1200/JCO.23.01423.

88. Hilal T., Sonbol M.B., Prasad V. Analysis of control arm quality in randomized clinical trials leading to anticancer drug approval by the US food and drug administration. JAMA Oncol. 2019;5(6):887–892. doi: 10.1001/jamaoncol.2019.0167.

89. Rossi A., Aimar G., Audisio M., et al. Analysis of the adequacy of control arms in oncology randomized clinical trials published between 2017 and 2021: a meta-research study. Eur J Cancer. 2023;189:112920. doi: 10.1016/j.ejca.2023.05.008. https://www.ejcancer.com/article/S0959-8049(23)00250-2/fulltext [cited 2023 May 18]. Available from:

90. Gyawali B., Kesselheim A.S. FDA approval standards for anticancer agents - lessons from two recent approvals in breast cancer. Nat Rev Clin Oncol. 2021;18(7):397–398. doi: 10.1038/s41571-021-00504-1.

91. Amit O., Bushnell W., Dodd L., Roach N., Sargent D. Blinded independent central review of the progression-free survival endpoint. Oncologist. 2010;15(5):492–495. doi: 10.1634/theoncologist.2009-0261.

92. Stone A., Gebski V., Davidson R., Bloomfield R., Bartlett J.W., Sabin A. Exaggeration of PFS by blinded, independent, central review (BICR) Ann Oncol. 2019;30(2):332–338. doi: 10.1093/annonc/mdy514.

93. Hwang T.J., Gyawali B. Association between progression-free survival and patients' quality of life in cancer clinical trials. Int J Cancer. 2019;144(7):1746–1751. doi: 10.1002/ijc.31957.

94. Kovic B., Jin X., Kennedy S.A., et al. Evaluating progression-free survival as a surrogate outcome for health-related quality of life in oncology: a systematic review and quantitative analysis. JAMA Intern Med. 2018;178(12):1586–1596. doi: 10.1001/jamainternmed.2018.4710.

95. Samuel J.N., Booth C.M., Eisenhauer E., Brundage M., Berry S.R., Gyawali B. Association of quality-of-life outcomes in cancer drug trials with survival outcomes and drug class. JAMA Oncol. 2022;8(6):879–886. doi: 10.1001/jamaoncol.2022.0864.

96. Sullivan H.W., O'Donoghue A.C., Boudewyns V., Paquin R.S., Ferriola-Bruckenstein K. Patient understanding of oncology clinical trial endpoints in direct-to-consumer television advertising. Oncologist. 2023;28(7):e542–e553. doi: 10.1093/oncolo/oyad064.

97. Gyawali B., Eisenhauer E., Tregear M., Booth C.M. Progression-free survival: it is time for a new name. Lancet Oncol. 2022;23(3):328–330. doi: 10.1016/S1470-2045(22)00015-8.

98. Brundage M.D., Booth C.M., Eisenhauer E.A., et al. Patients' attitudes and preferences toward delayed disease progression in the absence of improved survival. J Natl Cancer Inst. 2023;17 doi: 10.1093/jnci/djad138.

99. Rugo H.S., Im S.A., Cardoso F., et al. Margetuximab versus trastuzumab in patients with previously treated HER2-positive advanced breast cancer (SOPHIA): final overall survival results from a randomized phase 3 trial. J Clin Oncol. 2023;41(2):198–205. doi: 10.1200/JCO.21.02937.

100. Rugo H.S., Im S.A., Cardoso F., et al. Efficacy of margetuximab vs trastuzumab in patients with pretreated ERBB2-positive advanced breast cancer: a phase 3 randomized clinical trial. JAMA Oncol. 2021;7(4):573–584. doi: 10.1001/jamaoncol.2020.7932.

101. Saura C., Oliveira M., Feng Y.H., et al. Neratinib plus capecitabine versus lapatinib plus capecitabine in HER2-positive metastatic breast cancer previously treated with ≥ 2 HER2-directed regimens: phase III NALA trial. J Clin Oncol. 2020;38(27):3138–3149. doi: 10.1200/JCO.20.00147.

102. Hey S.P., Gyawali B., D'Andrea E., Kanagaraj M., Franklin J.M., Kesselheim A.S. A systematic review and meta-analysis of bevacizumab in first-line metastatic breast cancer: lessons for research and regulatory enterprises. J Natl Cancer Inst. 2019;112(4):335–342. doi: 10.1093/jnci/djz211.

103. Commissioner O of the. U.S. Food and Drug Administration . FDA; 2023. Clinical trial considerations to support accelerated approval of oncology therapeutics. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/clinical-trial-considerations-support-accelerated-approval-oncology-therapeutics [cited 2023 May 30] Available from:

104. Gyawali B., Kesselheim A.S., Ross J.S. The accelerated approval program for cancer drugs - finding the right balance. N Engl J Med. 2023;389(11):968–971. doi: 10.1056/NEJMp2306872.

105. Gyawali B., Prasad V. Combining drugs and extending treatment - a PFS end point is not sufficient. Nat Rev Clin Oncol. 2017;14(9):521–522. doi: 10.1038/nrclinonc.2017.72.

106. Gyawali B., de Vries, Dafni U., et al. Biases in study design, implementation, and data analysis that distort the appraisal of clinical benefit and ESMO-Magnitude of Clinical Benefit Scale (ESMO-MCBS) scoring. ESMO Open. 2021;6(3):100117. doi: 10.1016/j.esmoop.2021.100117.