Latest Cancer TreatmentsDr. Mohammadreza Dabiri
Dr. Mohammadreza Dabiriفارسی
Physician lesson

Critical reading of oncology trials and interpreting biomarkers

Endpoints, hazard ratio and confidence interval, non-inferiority design, crossover, subgroups, quality of life, ESMO-MCBS, NGS and ESCAT, tumour-agnostic approvals and uses of ctDNA; with real examples

Updated: 28 October 202610 min read51 references

Learning objectives and key messages

  • After this lesson you should be able to assess a trial's endpoint, translate a hazard ratio (HR) and confidence interval into relative and absolute effects, and read non-inferiority designs, crossover and subgroup claims critically [1, 2, 3, 4].
  • Overall survival is the most reliable endpoint; more than half of the reported correlations between surrogate endpoints and overall survival have been weak [1, 5].
  • Pathological complete response (pCR) is a strong prognostic factor at the patient level, but an increase in it within a trial does not by itself guarantee better survival [6].
  • In a non-inferiority trial, the whole confidence interval must lie on the right side of the prespecified margin; "non-inferiority not proven" is not the same as "worse" [3, 7].
  • ESMO-MCBS grades the magnitude of clinical benefit of a treatment (A and B, or 5 and 4 = substantial benefit), and ESCAT grades the evidence for the clinical actionability of a molecular target (levels I to X) [8, 9].
  • Tumour-agnostic approvals (MSI-H/dMMR, TMB-H, NTRK fusion, BRAF V600E, RET fusion, HER2 IHC ‎3+‎) rest mostly on objective response in single-arm basket trials, and response differs between tumour types [1, 10, 11].
  • ctDNA is in routine use for genotyping in advanced cancer (with reflex tissue testing after a negative result), and new randomised trials have shown its role in guiding adjuvant treatment and early treatment switching [12, 13, 14, 15].

Endpoints: what exactly was measured?

Common endpoints in oncology trials
EndpointDefinitionAdvantageLimitation
Overall survival (OS)Time from randomisation to death from any cause, in the intention-to-treat populationThe most reliable endpoint; precise and free of measurement bias [1]Long follow-up and a large sample; confounded by subsequent treatments [1]
Progression-free survival (PFS)Time from randomisation to objective tumour progression or death, whichever comes firstEarlier than OS and not confounded by subsequent treatment; a larger effect with a smaller sample [1]A precise definition of progression and blinded assessment are needed; data are usually insufficient to assess the correlation with OS [1]
Time to progression (TTP)Time to objective progression; deaths are not counted as events—Deaths are censored; PFS is preferred [1]
Disease-free survival (DFS)Time from randomisation to recurrence or death from any cause; mainly in adjuvant treatmentWhen survival is long and OS is impractical [1]Its value depends on the size of the effect, the toxicity of treatment and the disease setting [1]
Event-free survival (EFS)Randomisation before surgery; event = progression preventing surgery, local or distant recurrence or deathA suitable endpoint for neoadjuvant treatment [1]As for DFS
Objective response rate (ORR)The sum of partial and complete responses (stable disease is not included)A direct measure of antitumour activity; can be assessed in a single-arm trial [1]Its value depends on the size and durability of response and the proportion of complete responses; the commonest surrogate in accelerated approval [1]
Pathological complete response (pCR)Elimination of invasive tumour from the breast and lymph nodes (ypT0/is ypN0)Early; has been a surrogate for accelerated approval in high-risk early breast cancer [1, 6]At the trial level it correlated negligibly with EFS and OS [6]
Time to treatment failure (TTF)Time to stopping treatment for any reason (progression, toxicity or death)—Usually not recommended for approval of a new drug [1]
Symptoms and quality of life (patient-reported outcome)Measuring symptoms and function as reported by the patientImprovement of cancer-related symptoms is a direct measure of clinical benefit [1]Needs patients to be symptomatic at baseline; bias in open-label designs [1]

Definitions based on the FDA guidance (2018) [1].

Is the surrogate endpoint valid? In the systematic review by Prasad and colleagues, of 65 trial-level correlations between surrogate endpoints and overall survival (in 36 papers), 52% were weak (r ≤0.7), 25% moderate and 23% strong (r ≥0.85) [5]. In the CTNeoBC pooled analysis (11,955 patients with breast cancer), patients who achieved pCR (ypT0/is ypN0) had better survival (especially triple-negative disease: EFS HR 0.24 and OS HR 0.16), but at the trial level an increase in the pCR rate correlated negligibly with EFS (R² = 0.03) and OS (R² = 0.24), and pCR was not validated as a surrogate [6]. This is the difference between a "prognostic factor at the patient level" and a "valid surrogate at the trial level".

Worked example: KEYNOTE-522. Adding pembrolizumab to neoadjuvant chemotherapy in triple-negative breast cancer increased pCR from 51.2% to 64.8% (difference 13.6 percentage points; 95% CI 5.4 to 21.8) [16]; then 3-year EFS became 84.5% versus 76.8% (HR 0.63; 95% CI 0.48 to 0.82) [17], and finally 5-year overall survival was 86.6% versus 81.7% (P=0.002) [18]. In this particular case the surrogate benefit was confirmed by real outcomes, something CTNeoBC does not guarantee in general.

Quality of life. On the EORTC QLQ-C30 questionnaire, a change of 5 to 10 points corresponded to a "small" change, 10 to 20 to a "moderate" change and more than 20 to a "large" change from the patient's point of view [19]. In a randomised trial (766 patients), systematic monitoring of patient-reported symptoms improved quality of life, reduced emergency visits (34% vs 41%), and 1-year survival was 75% versus 69% (P=0.05) [20].

Hazard ratio, confidence interval and multiple testing

The hazard ratio (HR) is the ratio of the "rate" at which events occur in two groups and assumes that this ratio is constant over time; when this assumption does not hold, the clinical interpretation of the HR is difficult or impossible, and Uno and colleagues propose reporting a model-free, prespecified and clinically meaningful measure of the difference between the two groups [2].

Relative versus absolute effect. An HR of 0.63 for EFS in KEYNOTE-522 means a relative reduction of about 37% in the instantaneous risk of an event, while the absolute difference in 3-year EFS is 7.7 percentage points (84.5% vs 76.8%); that is, about 13 patients must be treated to prevent one event in 3 years (1 divided by 0.077) [17]. Always look at both together.

Confidence interval and the significance boundary. In the final overall survival analysis of FLAURA (osimertinib versus a comparator EGFR TKI in EGFR-mutated lung cancer), median survival was 38.6 versus 31.8 months with an HR of 0.80 and a 95.05% confidence interval of 0.64 to 1.00 (P=0.046) [21]; the upper limit of the confidence interval reached 1, and the true size of the effect carries considerable uncertainty. Note also the unusual confidence levels (95.05% in FLAURA and 98.4% for OS in CheckMate 649) [21, 22]; these levels come from the prespecified statistical plan for several endpoints and interim analyses, and the result must be interpreted at that level.

Hierarchical testing. In CheckMate 649, the primary endpoints were OS and PFS in patients with PD-L1 CPS ≥5, and OS in CPS ≥1 and in all patients was tested hierarchically [22, 23]; so a significant result in "all patients" does not mean the same benefit in every subgroup (see the section on crossover, subsequent treatments and subgroups).

Non-inferiority trials

A non-inferiority trial does not show that a new treatment is "not worse"; it shows that it is not worse by more than a small prespecified amount (the non-inferiority margin, or Δ). According to the EMA guideline, the margin must be prespecified in the protocol and justified both clinically and statistically, and at the end of the study the two-sided 95% (or one-sided 97.5%) confidence interval of the difference must lie entirely on the positive side of the margin [3]. The margin should not be set as a percentage of the comparator's superiority over placebo or based on the study's power (a small study is no justification for a wider margin); and if a result lying between −Δ and zero does not seem clinically acceptable, Δ was not chosen correctly [3]. A CONSORT extension has been developed for standard reporting of non-inferiority and equivalence trials [24].

Four real examples of non-inferiority design
TrialQuestionNon-inferiority marginResultLesson
FAST-ForwardRadiotherapy of 26 or 27 Gy in 5 fractions (1 week) versus 40 Gy in 15 fractions after breast cancer surgeryAt most a 1.6% increase in 5-year local relapse (critical HR 1.81)5-year relapse with 40 Gy 2.1%; absolute difference for 26 Gy −0.7% (−1.3 to 0.3) [25]Non-inferiority of 26 Gy was proven and its normal-tissue effects were similar to 40 Gy, but 27 Gy had more side effects [25]
IDEA3 versus 6 months of adjuvant chemotherapy (FOLFOX or CAPOX) in stage 3 colon cancerUpper limit of the two-sided 95% confidence interval of the HR at most 1.12Whole population HR 1.07 (1.00 to 1.15); CAPOX 0.95 (0.85 to 1.06); FOLFOX 1.16 (1.06 to 1.26) [26]Non-inferiority was not proven in the whole population but was for CAPOX; the analysis by risk (T1–3 N1) was exploratory [26]
DYNAMICctDNA-guided management versus clinicopathological criteria in stage 2 colon cancer−8.5 percentage points in 2-year recurrence-free survival93.5% versus 92.4% (difference 1.1; 95% CI −4.1 to 6.2) with less chemotherapy (15% vs 28%) [13]Non-inferiority was proven; the critical question: is a margin of 8.5 percentage points acceptable to patients?
SCORADSingle-fraction 8 Gy versus 20 Gy in 5 fractions in metastatic spinal cord compression−11% in the proportion of patients able to walk at week 869.3% versus 72.7% (difference −3.5%; one-sided lower limit −11.5%) [7]Non-inferiority was not proven, but the lower limit only just crossed the margin; "not proven" is not the same as "worse" [7]

Note the direction of the margin: in IDEA the margin is defined on the HR (upper limit), and in DYNAMIC and SCORAD on the absolute difference (lower limit).

Crossover, subsequent treatments and subgroups

Crossover and subsequent treatments. When control patients move to the experimental treatment after progression, a smaller (and sometimes zero) OS difference is seen. Korn and colleagues regard a crossover design in the first definitive evaluation of a new drug as interfering with the measurement of clinical benefit, and recommend that the observed OS difference, provided subsequent treatments in both arms follow the standard of care, be taken as the measure of clinical benefit [27]. In PROfound (olaparib in castration-resistant prostate cancer with BRCA1, BRCA2 or ATM mutations), 66% of control patients crossed over to olaparib; even so, OS in cohort A was significant (19.1 vs 14.7 months; HR 0.69; 95% CI 0.50 to 0.97), and a sensitivity analysis adjusted for crossover estimated the HR at 0.42 [28]. In KEYNOTE-024 crossover was also allowed, and 6-month survival was better with pembrolizumab (80.2% vs 72.4%; HR 0.60) [29].

Subgroups. The EMA guideline makes the credibility of a subgroup finding depend on a prespecified and well-argued definition, biological rationale and replication in independent data; examining many subgroups increases the chance of a false-positive finding, and the balance achieved by randomisation may not hold within subgroups. Interaction tests have low power, and a single P value for an interaction test is not a sufficient basis for decisions; the effect estimate with its confidence interval, together with the complementary subgroup, should be presented [4]. The ICEMAN tool also proposes 5 core questions for judging the credibility of subgroup effect claims in trials [30].

Worked example: PD-L1 in CheckMate 649. At 3-year follow-up, the unstratified HR for overall survival with nivolumab + chemotherapy in CPS ≥1, ≥5 and ≥10 was 0.75, 0.69 and 0.66, but in CPS below 1, below 5 and below 10 it was 0.95, 0.95 and 0.91; the authors judged the magnitude of benefit "enriched" at CPS ≥5 and the extra benefit at CPS ≥10 negligible [23]. So a significant benefit in "all patients" should not be extended without qualification to a patient with a low CPS, although these analyses are exploratory [4, 23].

Clinical value: ESMO-MCBS

The ESMO Magnitude of Clinical Benefit Scale (ESMO-MCBS) is a validated and reproducible tool for grading the magnitude of clinical benefit of new anticancer treatments; version 1.0 was published in 2015 [31]. Scores run from A to C for curative treatment and from 5 to 1 for non-curative treatment, and scores of A and B, or 5 and 4, count as "substantial benefit" [8].

  • Version 1.1 (2017): 10 amendments, including the ability to score single-arm trials; the score changed for only 12 of 118 comparative studies [32].
  • Version 2.0 (2025): 13 amendments; in curative treatment, longer time free of treatment or disease earns credit even without a significant OS improvement, small absolute DFS gains are scored more cautiously, and a label for acute and lasting toxicity has been added; in non-curative treatment, "tail of the curve" benefit in OS and PFS is assessed with a more robust statistical method and the toxicity assessment has been revised; these changes altered the score of 13.6% of studies (10.5% down, 3.1% up) [33].
  • ESMO-MCBS:H (2023): a version for haematological malignancies, with 5 amendments for endpoints specific to haematology and for indolent but incurable diseases such as follicular lymphoma [34].

In the EHA assessment with version 1.1, CAR-T cell therapy in relapsed ALL in children and young adults received at most a score of 3 because the study was single-arm; an example of how study design sets the ceiling of the score [8].

NGS, ESCAT and tumour-agnostic approvals

ESCAT. The ESMO Scale for Clinical Actionability of molecular Targets defines six levels of evidence: level I, targets ready for routine decisions; level II, investigational targets that probably define a population benefiting from a targeted drug but need more data; level III, clinical benefit proven in other tumour types or for similar molecular targets; level IV, preclinical evidence; level V, evidence supporting co-targeting; and level X, no evidence of actionability [9].

ESMO 2024 NGS recommendations. Tumour NGS is recommended in advanced non-squamous non-small cell lung cancer, prostate, colorectal, cholangiocarcinoma and ovarian cancer, and in this update it has been extended to advanced breast cancer and rare tumours (GIST, sarcoma, thyroid and cancer of unknown primary); it is also recommended for detecting tumour-agnostic alterations in metastatic cancers when a matched treatment is accessible [35].

The rationale and limits of tumour-agnostic approval. These approvals are often based on single-arm basket trials with objective response and duration of response as endpoints; ORR is a direct measure of antitumour activity and the commonest surrogate endpoint for accelerated approval [1]. But "tumour-agnostic" does not mean "regardless of tumour type": in the ROAR basket trial with dabrafenib + trametinib in BRAF V600E mutation, response ranged from 89% in hairy cell leukaemia and 56% in anaplastic thyroid carcinoma to 33% in high-grade glioma [10]; and in DESTINY-PanTumor02, response to trastuzumab deruxtecan in centrally confirmed HER2 IHC ‎3+‎ was 61.3% versus 37.1% in all patients (IHC ‎3+‎ or ‎2+‎) [11].

Tumour-agnostic approvals (FDA labels, updated 2026)
Molecular alterationDrugKey evidence (objective response)Status in the FDA label
MSI-H or dMMRPembrolizumabKEYNOTE-158, previously treated non-colorectal tumours (27 tumour types): 34.3% (95% CI 28.3 to 40.8) [36]Approved for advanced MSI-H/dMMR solid tumours after previous treatment; no accelerated-approval qualifier in the current label [37]
dMMRDostarlimabGARNET, 327 patients: 44.0% (95% CI 38.6 to 49.6) [38]Accelerated approval (based on response and duration of response) [39]
TMB-H (≥10 mutations per megabase)PembrolizumabKEYNOTE-158: 29% in high TMB versus 6% in low TMB [40]Accelerated approval; safety and efficacy in paediatric CNS tumours with TMB-H have not been established [37]
NTRK fusionLarotrectinib75% by independent review in the first 55 patients with 17 tumour types [41]; 79% in a pooled analysis of 153 patients [42]Approved for solid tumours with NTRK fusion without a known acquired resistance mutation; no accelerated-approval qualifier in the current label [43]
NTRK fusionEntrectinib57% in 54 adults [44]; 61.2% in 121 patients, with intracranial response of 63.6% [45]Accelerated approval [46]
NTRK fusionRepotrectinibTRIDENT-1: 58% in TKI-naive patients and 50% after a previous TKI [47]Accelerated approval (age 12 and over) [47]
BRAF V600EDabrafenib + trametinibROAR: response from 33% to 89% depending on tumour type [10]Accelerated approval (age 1 and over) [48]
RET fusionSelpercatinibLIBRETTO-001 (other than lung and thyroid): 43.9% (95% CI 28.5 to 60.3) [49]Approved for solid tumours with RET fusion (age 2 and over); no accelerated-approval qualifier in the current label [50]
HER2 IHC ‎3+‎Trastuzumab deruxtecanDESTINY-PanTumor02: 61.3% in centrally confirmed IHC ‎3+‎ [11]Accelerated approval [51]

All indications are for advanced tumours after previous treatment or with no satisfactory alternative; see each drug's label for details and the approved diagnostic test.

Circulating tumour DNA (ctDNA)

ESMO guideline (2022). In advanced cancer, validated ctDNA tests with adequate sensitivity can be used in routine practice to detect actionable mutations, but because of limitations in detecting fusions and copy number changes, tissue testing is preferred for many patients; ctDNA is used when speed of results matters or when a biopsy is not possible or appropriate, and reflex tissue testing is recommended after an uninformative ctDNA result. In early cancer, detection of molecular residual disease (MRD) has high clinical validity for predicting relapse, but when the guideline was published it was not recommended for routine practice because of a lack of evidence of clinical utility in guiding treatment [12].

Newer randomised evidence.

  • DYNAMIC: in stage 2 colon cancer, ctDNA-guided management reduced the use of adjuvant chemotherapy (15% vs 28%) without reducing 2-year recurrence-free survival [13].
  • IMvigor011: in muscle-invasive bladder cancer with positive ctDNA after cystectomy, adjuvant atezolizumab improved disease-free survival (HR 0.64) and overall survival (HR 0.59) compared with placebo; in 357 patients with persistently negative ctDNA and no treatment, 2-year disease-free survival was 88% [14].
  • SERENA-6: in advanced ER-positive, HER2-negative breast cancer on an aromatase inhibitor and a CDK4/6 inhibitor, an early switch to camizestrant when an ESR1 mutation emerged in ctDNA (before radiological progression) increased PFS from 9.2 to 16.0 months (HR 0.44) [15]; the primary endpoint was investigator-assessed PFS at an interim analysis, and any overall survival benefit must be assessed separately.

For a patient-centred review of this topic, see the article on liquid biopsy and ctDNA.

Step-by-step checklist for appraising an oncology trial

  1. 1

    1. Question and population

    Define the population (inclusion criteria, line of treatment, biomarker), intervention, comparator and outcome, and ask whether your patient resembles the study patients.

  2. 2

    2. Endpoint

    Is it OS or a surrogate? For PFS and ORR, was there blinded independent assessment? Has the surrogate been validated in this setting [1, 5]?

  3. 3

    3. Effect size

    Read the HR with its confidence interval, the absolute difference and the NNT, and check whether the proportional hazards assumption holds [2].

  4. 4

    4. Statistical design

    Superiority or non-inferiority? A prespecified margin and its justification? Hierarchical testing and the alpha level of each analysis [3, 22].

  5. 5

    5. Subsequent treatments and crossover

    Were post-progression treatments standard in both arms? How much has crossover diluted the OS difference [27]?

  6. 6

    6. Subgroups

    Prespecified, biologically justified and replicated? What is the effect estimate in the complementary subgroup [4]?

  7. 7

    7. Toxicity and quality of life

    Look at grade 3 or higher side effects, treatment discontinuation and clinically meaningful changes in quality of life (about 10 points or more on the QLQ-C30) [19].

  8. 8

    8. Value and applicability

    Include the ESMO-MCBS score, the ESCAT level of the molecular target and the patient's actual access to the treatment in the decision [9, 33].

Practical and exam points

  • Overall survival is the most reliable endpoint; PFS is not confounded by subsequent treatments, but data are usually insufficient to assess its correlation with OS [1].
  • ORR includes complete + partial responses, not stable disease [1].
  • 52% of trial-level surrogate–OS correlations were weak [5]; pCR is not a valid trial-level surrogate [6].
  • NNT = 1 divided by the absolute difference; in KEYNOTE-522 about 13 for 3-year EFS [17].
  • Non-inferiority: the whole confidence interval on the right side of the margin; IDEA, with an upper limit of 1.15 and a margin of 1.12, did not show overall non-inferiority [3, 26].
  • Crossover dilutes the OS difference; the observed difference with standard subsequent treatments is the measure of clinical benefit [27].
  • The P value of an interaction test alone is not enough; look at the effect in the complementary subgroup too [4].
  • ESMO-MCBS: A and B (curative) or 5 and 4 (non-curative) = substantial benefit [8].
  • ESCAT: level I ready for routine use, level X no evidence of actionability [9].
  • A negative ctDNA in advanced disease does not exclude a mutation; do reflex tissue testing [12].

Studies and guidelines used as examples

Studies and guidelines used as examples

15 items
Clinical guideline

Recommendations for the use of next-generation sequencing (NGS) for patients with advanced cancer in 2024: a report from the ESMO Precision Medicine Working Group

Annals of Oncology · 2024
Advanced non-squamous non-small-cell lung cancer, prostate cancer, colorectal cancer, cholangiocarcinoma and ovarian cancerCancers for which NGS was already recommended in 2020
Summary
Clinical trial · Phase 3

Overall Survival with Pembrolizumab in Early-Stage Triple-Negative Breast Cancer

N Engl J Med · 2024
86.6% vs 81.7% (pembrolizumab–chemotherapy vs placebo–chemotherapy)Estimated overall survival at 60 months
Summary
Clinical trial · Phase 3

Pembrolizumab versus Chemotherapy for PD-L1–Positive Non–Small-Cell Lung Cancer

N Engl J Med · 2016
10.3 months vs 6.0 monthsMedian progression-free survival
Summary
Clinical trial · Phase 3

Overall Survival with Osimertinib in Untreated, EGFR -Mutated Advanced NSCLC

N Engl J Med · 2020
38.6 months vs 31.8 monthsMedian overall survival
Summary
Clinical trial · Phase 3

Survival with Olaparib in Metastatic Castration-Resistant Prostate Cancer

N Engl J Med · 2020
19.1 months with olaparib vs 14.7 months with control treatmentMedian overall survival in cohort A (BRCA1, BRCA2 or ATM)
Summary
Clinical trial · Phase 2

Efficacy and Safety of Trastuzumab Deruxtecan in Patients With HER2-Expressing Solid Tumors: Primary Results From the DESTINY-PanTumor02 Phase II Trial

J Clin Oncol · 2024
37.1%Objective response in all patients
Summary
Clinical trial · Phase 3

Hypofractionated breast radiotherapy for 1 week versus 3 weeks (FAST-Forward): 5-year efficacy and late normal tissue effects results from a multicentre, non-inferiority, randomised, phase 3 trial

The Lancet · 2020
31 patients (40 Gy), 27 patients (27 Gy), 21 patients (26 Gy)Ipsilateral breast tumour relapse (number of events)
Summary
Clinical trial

Duration of Adjuvant Chemotherapy for Stage III Colon Cancer

N Engl J Med · 2018
HR 1.07Overall population (3 vs 6 months)
Summary
Clinical trial

Effect of Single-Fraction vs Multifraction Radiotherapy on Ambulatory Status Among Patients With Spinal Canal Compression From Metastatic Cancer: The SCORAD Randomized Clinical Trial

JAMA · 2019
69.3% vs 72.7%Ambulatory (grade 1 or 2) at week 8 — primary outcome
Summary
Clinical trial · Phase 2

Circulating Tumor DNA Analysis Guiding Adjuvant Therapy in Stage II Colon Cancer

N Engl J Med · 2022
15% vs 28%Receipt of adjuvant chemotherapy
Summary
Clinical trial · Phase 2

Mismatch repair deficiency predicts response of solid tumors to PD-1 blockade

Science · 2017
53% of patientsObjective radiographic response
Summary
Clinical trial · Phase 3

ctDNA-Guided Adjuvant Atezolizumab in Muscle-Invasive Bladder Cancer

N Engl J Med · 2025
9.9 vs 4.8 monthsMedian disease-free survival
Summary
Clinical trial · Phase 3

First-Line Camizestrant for Emerging ESR1-Mutated Advanced Breast Cancer

N Engl J Med · 2025
16.0 vs 9.2 monthsMedian progression-free survival
Summary
Clinical trial · Phase 3

Pembrolizumab for Early Triple-Negative Breast Cancer

N Engl J Med · 2020
64.8% vs 51.2%Pathological complete response
Summary
Clinical trial · Phase 3

First-line nivolumab plus chemotherapy versus chemotherapy alone for advanced gastric, gastro-oesophageal junction, and oesophageal adenocarcinoma (CheckMate 649): a randomised, open-label, phase 3 trial

Lancet · 2021
14.4 vs 11.1 monthsMedian overall survival (CPS ≥5)
Summary

Self-assessment questions

In KEYNOTE-522, the HR for event-free survival was 0.63 and 3-year EFS was 84.5% versus 76.8%. Calculate the relative risk reduction, the absolute difference and the NNT.

Relative reduction in instantaneous risk about 37% (1 minus 0.63); absolute difference 7.7 percentage points; NNT over 3 years about 13 (1 divided by 0.077) [17].

Why does an increase in pCR in a trial not guarantee better survival?

In CTNeoBC, pCR at the patient level was associated with better survival, but at the trial level an increase in pCR predicted almost nothing about improvement in EFS (R² = 0.03) or OS (R² = 0.24) [6]; most trial-level surrogate–OS correlations in oncology are also weak [5].

In IDEA, the HR for the whole population was 1.07 with a confidence interval of 1.00 to 1.15 and a non-inferiority margin of 1.12. What is the conclusion and its clinical application?

Because the upper limit of the confidence interval (1.15) is above the margin (1.12), non-inferiority of 3 months was not proven in the whole population; but non-inferiority was shown for CAPOX (0.95; 0.85 to 1.06) and not for FOLFOX (1.16; 1.06 to 1.26). In an exploratory analysis, 3 months of treatment was non-inferior in T1–3 N1 and 6 months was superior in T4 or N2 [26].

In PROfound, OS was significant despite 66% crossover. How do you interpret the crossover-adjusted HR?

The observed HR in cohort A (0.69; 0.50 to 0.97) shows the real benefit under trial conditions, despite the control group's access to olaparib; the adjusted HR (0.42) is a sensitivity analysis estimating the effect in the absence of crossover [28]. Korn and colleagues regard the observed difference, provided subsequent treatments are standard, as the main measure of clinical benefit [27].

In CheckMate 649, OS in the whole population was significant. Do you expect the same benefit for a patient with PD-L1 CPS below 1?

Not necessarily; the unstratified HR was 0.95 at CPS below 1 and 0.69 at CPS ≥5 [23]. This subgroup comparison is exploratory and should be judged with the complementary subgroup, the confidence interval and biological rationale, not only the P value of the interaction test [4].

What is the difference between ESCAT and ESMO-MCBS?

ESCAT ranks a molecular target by the evidence for its clinical actionability (levels I to X) [9]; ESMO-MCBS grades the magnitude of clinical benefit of a treatment from trial data (A to C for curative and 5 to 1 for non-curative treatment) [8, 31].

NGS in a patient with previously treated metastatic cancer reports an NTRK fusion. What is the evidence for treatment, and why does it come mostly from single-arm trials?

Larotrectinib had a response of 75% (independent review) in 55 patients with 17 tumour types [41], and entrectinib a response of 57% in 54 adults [44], and both are approved in FDA labels for solid tumours with NTRK fusion [43, 46]. This evidence comes from single-arm basket trials with an ORR endpoint; ORR can be assessed in a single-arm design and is the commonest surrogate for accelerated approval [1].

The ctDNA of a patient with advanced lung cancer is reported as negative. Can we say there is no actionable mutation or fusion?

No; ctDNA tests have false-negative results and are limited in detecting fusions and copy number changes; after an uninformative result, reflex tissue testing is recommended [12].

References

51 references
Show all references, 14 with a study cardClose the reference list

Numbers in the text, such as [1], refer to these sources. Sources with a “Study card” button are summarised in the site’s evidence library.

  1. 1
    Clinical Trial Endpoints for the Approval of Cancer Drugs and Biologics: Guidance for IndustryU.S. Food and Drug Administration · fda.gov · 2018
    Source Back to text: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
  2. 2
    Moving beyond the hazard ratio in quantifying the between-group difference in survival analysisUno H, Claggett B, Tian L, Inoue E, Gallo P, Miyata T, et al · J Clin Oncol · 2014;32:2380-5
    DOIPubMed Back to text: 1 2 3
  3. 3
    Guideline on the choice of the non-inferiority margin (EMEA/CPMP/EWP/2158/99)European Medicines Agency, Committee for Medicinal Products for Human Use (CHMP) · ema.europa.eu · 2005
    Source Back to text: 1 2 3 4 5 6
  4. 4
    Guideline on the investigation of subgroups in confirmatory clinical trials (EMA/CHMP/539146/2013)European Medicines Agency, Committee for Medicinal Products for Human Use (CHMP) · ema.europa.eu · 2019
    Source Back to text: 1 2 3 4 5 6
  5. 5
    The Strength of Association Between Surrogate End Points and Survival in Oncology: A Systematic Review of Trial-Level Meta-analysesPrasad V, Kim C, Burotto M, Vandross A · JAMA Intern Med · 2015;175:1389-98
    DOIPubMed Back to text: 1 2 3 4 5
  6. 6
    Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysisCortazar P, Zhang L, Untch M, Mehta K, Costantino JP, Wolmark N, et al · Lancet · 2014;384:164-72
    DOIPubMed Back to text: 1 2 3 4 5 6
  7. 7
    Effect of Single-Fraction vs Multifraction Radiotherapy on Ambulatory Status Among Patients With Spinal Canal Compression From Metastatic Cancer: The SCORAD Randomized Clinical TrialHoskin PJ, Hopkins K, Misra V, Holt T, McMenemin R, Dubois D, et al · JAMA · 2019;322:2084-2094
    Study cardDOIPubMed Back to text: 1 2 3
  8. 8
    EHA evaluation of the ESMO-Magnitude of Clinical Benefit Scale version 1.1 (ESMO-MCBS v1.1) for haematological malignanciesKiesewetter B, Cherny NI, Boissel N, Cerisoli F, Dafni U, de Vries EGE, et al · ESMO Open · 2020;5:e000611
    DOIPubMed Back to text: 1 2 3 4 5
  9. 9
    A framework to rank genomic alterations as targets for cancer precision medicine: the ESMO Scale for Clinical Actionability of molecular Targets (ESCAT)Mateo J, Chakravarty D, Dienstmann R, Jezdic S, Gonzalez-Perez A, Lopez-Bigas N, et al · Ann Oncol · 2018;29:1895-1902
    DOIPubMed Back to text: 1 2 3 4 5
  10. 10
    Dabrafenib plus trametinib in BRAFV600E-mutated rare cancers: the phase 2 ROAR trialSubbiah V, Kreitman RJ, Wainberg ZA, Gazzah A, Lassen U, Stein A, et al · Nat Med · 2023;29:1103-1112
    DOIPubMed Back to text: 1 2 3
  11. 11
    Efficacy and Safety of Trastuzumab Deruxtecan in Patients With HER2-Expressing Solid Tumors: Primary Results From the DESTINY-PanTumor02 Phase II TrialMeric-Bernstam F, Makker V, Oaknin A, Oh DY, Banerjee S, González-Martín A, et al · J Clin Oncol · 2024;42:47-58
    Study cardDOIPubMed Back to text: 1 2 3
  12. 12
    ESMO recommendations on the use of circulating tumour DNA assays for patients with cancer: a report from the ESMO Precision Medicine Working GroupPascual J, Attard G, Bidard FC, Curigliano G, De Mattos-Arruda L, Diehn M, et al · Ann Oncol · 2022;33:750-768
    DOIPubMed Back to text: 1 2 3 4
  13. 13
    Circulating Tumor DNA Analysis Guiding Adjuvant Therapy in Stage II Colon CancerTie J, Cohen JD, Lahouel K, Lo SN, Wang Y, Kosmider S, et al · N Engl J Med · 2022;386:2261-2272
    Study cardDOIPubMed Back to text: 1 2 3
  14. 14
    ctDNA-Guided Adjuvant Atezolizumab in Muscle-Invasive Bladder CancerPowles T, Kann AG, Castellano D, Gross-Goupil M, Nishiyama H, Bracarda S, et al · N Engl J Med · 2025;393:2395-2408
    Study cardDOIPubMed Back to text: 1 2
  15. 15
    First-Line Camizestrant for Emerging ESR1-Mutated Advanced Breast CancerBidard FC, Mayer EL, Park YH, Janni W, Ma C, Cristofanilli M, et al · N Engl J Med · 2025;393:569-580
    Study cardDOIPubMed Back to text: 1 2
  16. 16
    Pembrolizumab for Early Triple-Negative Breast CancerSchmid P, Cortes J, Pusztai L, McArthur H, Kümmel S, Bergh J, et al · N Engl J Med · 2020;382:810-821
    Study cardDOIPubMed Back to text
  17. 17
    Event-free Survival with Pembrolizumab in Early Triple-Negative Breast CancerSchmid P, Cortes J, Dent R, Pusztai L, McArthur H, Kümmel S, et al · N Engl J Med · 2022;386:556-567
    DOIPubMed Back to text: 1 2 3 4
  18. 18
    Overall Survival with Pembrolizumab in Early-Stage Triple-Negative Breast CancerSchmid P, Cortes J, Dent R, McArthur H, Pusztai L, Kümmel S, et al · N Engl J Med · 2024;391:1981-1991
    Study cardDOIPubMed Back to text
  19. 19
    Interpreting the significance of changes in health-related quality-of-life scoresOsoba D, Rodrigues G, Myles J, Zee B, Pater J · J Clin Oncol · 1998;16:139-44
    DOIPubMed Back to text: 1 2
  20. 20
    Symptom Monitoring With Patient-Reported Outcomes During Routine Cancer Treatment: A Randomized Controlled TrialBasch E, Deal AM, Kris MG, Scher HI, Hudis CA, Sabbatini P, et al · J Clin Oncol · 2016;34:557-65
    DOIPubMed Back to text
  21. 21
    Overall Survival with Osimertinib in Untreated, EGFR-Mutated Advanced NSCLCRamalingam SS, Vansteenkiste J, Planchard D, Cho BC, Gray JE, Ohe Y, et al · N Engl J Med · 2020;382:41-50
    Study cardDOIPubMed Back to text: 1 2
  22. 22
    First-line nivolumab plus chemotherapy versus chemotherapy alone for advanced gastric, gastro-oesophageal junction, and oesophageal adenocarcinoma (CheckMate 649): a randomised, open-label, phase 3 trialJanjigian YY, Shitara K, Moehler M, Garrido M, Salman P, Shen L, et al · Lancet · 2021;398:27-40
    Study cardDOIPubMed Back to text: 1 2 3
  23. 23
    First-Line Nivolumab Plus Chemotherapy for Advanced Gastric, Gastroesophageal Junction, and Esophageal Adenocarcinoma: 3-Year Follow-Up of the Phase III CheckMate 649 TrialJanjigian YY, Ajani JA, Moehler M, Shen L, Garrido M, Gallardo C, et al · J Clin Oncol · 2024;42:2012-2020
    DOIPubMed Back to text: 1 2 3 4
  24. 24
    Reporting of noninferiority and equivalence randomized trials: extension of the CONSORT 2010 statementPiaggio G, Elbourne DR, Pocock SJ, Evans SJ, Altman DG · JAMA · 2012;308:2594-604
    DOIPubMed Back to text
  25. 25
    Hypofractionated breast radiotherapy for 1 week versus 3 weeks (FAST-Forward): 5-year efficacy and late normal tissue effects results from a multicentre, non-inferiority, randomised, phase 3 trialMurray Brunt A, Haviland JS, Wheatley DA, Sydenham MA, Alhasso A, Bloomfield DJ, et al · Lancet · 2020;395:1613-1626
    Study cardDOIPubMed Back to text: 1 2
  26. 26
    Duration of Adjuvant Chemotherapy for Stage III Colon CancerGrothey A, Sobrero AF, Shields AF, Yoshino T, Paul J, Taieb J, et al · N Engl J Med · 2018;378:1177-1188
    Study cardDOIPubMed Back to text: 1 2 3 4
  27. 27
    Overall survival as the outcome for randomized clinical trials with effective subsequent therapiesKorn EL, Freidlin B, Abrams JS · J Clin Oncol · 2011;29:2439-42
    DOIPubMed Back to text: 1 2 3 4
  28. 28
    Survival with Olaparib in Metastatic Castration-Resistant Prostate CancerHussain M, Mateo J, Fizazi K, Saad F, Shore N, Sandhu S, et al · N Engl J Med · 2020;383:2345-2357
    Study cardDOIPubMed Back to text: 1 2
  29. 29
    Pembrolizumab versus Chemotherapy for PD-L1-Positive Non-Small-Cell Lung CancerReck M, Rodríguez-Abreu D, Robinson AG, Hui R, Csőszi T, Fülöp A, et al · N Engl J Med · 2016;375:1823-1833
    Study cardDOIPubMed Back to text
  30. 30
    Development of the Instrument to assess the Credibility of Effect Modification Analyses (ICEMAN) in randomized controlled trials and meta-analysesSchandelmaier S, Briel M, Varadhan R, Schmid CH, Devasenapathy N, Hayward RA, et al · CMAJ · 2020;192:E901-E906
    DOIPubMed Back to text
  31. 31
    A standardised, generic, validated approach to stratify the magnitude of clinical benefit that can be anticipated from anti-cancer therapies: the European Society for Medical Oncology Magnitude of Clinical Benefit Scale (ESMO-MCBS)Cherny NI, Sullivan R, Dafni U, Kerst JM, Sobrero A, Zielinski C, et al · Ann Oncol · 2015;26:1547-73
    DOIPubMed Back to text: 1 2
  32. 32
    ESMO-Magnitude of Clinical Benefit Scale version 1.1Cherny NI, Dafni U, Bogaerts J, Latino NJ, Pentheroudakis G, Douillard JY, et al · Ann Oncol · 2017;28:2340-2366
    DOIPubMed Back to text
  33. 33
    ESMO-Magnitude of Clinical Benefit Scale version 2.0 (ESMO-MCBS v2.0)Cherny NI, Oosting SF, Dafni U, Latino NJ, Galotti M, Zygoura P, et al · Ann Oncol · 2025;36:866-908
    DOIPubMed Back to text: 1 2
  34. 34
    ESMO-Magnitude of Clinical Benefit Scale for haematological malignancies (ESMO-MCBS:H) version 1.0Kiesewetter B, Dafni U, de Vries EGE, Barriuso J, Curigliano G, González-Calle V, et al · Ann Oncol · 2023;34:734-771
    DOIPubMed Back to text
  35. 35
    Recommendations for the use of next-generation sequencing (NGS) for patients with advanced cancer in 2024: a report from the ESMO Precision Medicine Working GroupMosele MF, Westphalen CB, Stenzinger A, Barlesi F, Bayle A, Bièche I, et al · Ann Oncol · 2024;35:588-606
    Study cardDOIPubMed Back to text
  36. 36
    Efficacy of Pembrolizumab in Patients With Noncolorectal High Microsatellite Instability/Mismatch Repair-Deficient Cancer: Results From the Phase II KEYNOTE-158 StudyMarabelle A, Le DT, Ascierto PA, Di Giacomo AM, De Jesus-Acosta A, Delord JP, et al · J Clin Oncol · 2020;38:1-10
    DOIPubMed Back to text
  37. 37
    KEYTRUDA (pembrolizumab) injection, for intravenous use: Prescribing InformationMerck Sharp & Dohme LLC · DailyMed, U.S. National Library of Medicine · 2026
    Source Back to text: 1 2
  38. 38
    Antitumor Activity and Safety of Dostarlimab Monotherapy in Patients With Mismatch Repair Deficient Solid Tumors: A Nonrandomized Controlled TrialAndré T, Berton D, Curigliano G, Sabatier R, Tinker AV, Oaknin A, et al · JAMA Netw Open · 2023;6:e2341165
    DOIPubMed Back to text
  39. 39
    JEMPERLI (dostarlimab-gxly) injection, for intravenous use: Prescribing InformationGlaxoSmithKline LLC · DailyMed, U.S. National Library of Medicine · 2025
    Source Back to text
  40. 40
    Association of tumour mutational burden with outcomes in patients with advanced solid tumours treated with pembrolizumab: prospective biomarker analysis of the multicohort, open-label, phase 2 KEYNOTE-158 studyMarabelle A, Fakih M, Lopez J, Shah M, Shapira-Frommer R, Nakagawa K, et al · Lancet Oncol · 2020;21:1353-1365
    DOIPubMed Back to text
  41. 41
    Efficacy of Larotrectinib in TRK Fusion-Positive Cancers in Adults and ChildrenDrilon A, Laetsch TW, Kummar S, DuBois SG, Lassen UN, Demetri GD, et al · N Engl J Med · 2018;378:731-739
    DOIPubMed Back to text: 1 2
  42. 42
    Larotrectinib in patients with TRK fusion-positive solid tumours: a pooled analysis of three phase 1/2 clinical trialsHong DS, DuBois SG, Kummar S, Farago AF, Albert CM, Rohrberg KS, et al · Lancet Oncol · 2020;21:531-540
    DOIPubMed Back to text
  43. 43
    VITRAKVI (larotrectinib) capsules / oral solution: Prescribing InformationBayer HealthCare Pharmaceuticals Inc · DailyMed, U.S. National Library of Medicine · 2026
    Source Back to text: 1 2
  44. 44
    Entrectinib in patients with advanced or metastatic NTRK fusion-positive solid tumours: integrated analysis of three phase 1-2 trialsDoebele RC, Drilon A, Paz-Ares L, Siena S, Shaw AT, Farago AF, et al · Lancet Oncol · 2020;21:271-282
    DOIPubMed Back to text: 1 2
  45. 45
    Updated Integrated Analysis of the Efficacy and Safety of Entrectinib in Patients With NTRK Fusion-Positive Solid TumorsDemetri GD, De Braud F, Drilon A, Siena S, Patel MR, Cho BC, et al · Clin Cancer Res · 2022;28:1302-1312
    DOIPubMed Back to text
  46. 46
    ROZLYTREK (entrectinib) capsules / oral pellets: Prescribing InformationGenentech, Inc · DailyMed, U.S. National Library of Medicine · 2026
    Source Back to text: 1 2
  47. 47
    AUGTYRO (repotrectinib) capsules, for oral use: Prescribing InformationE.R. Squibb & Sons, L.L.C · DailyMed, U.S. National Library of Medicine · 2024
    Source Back to text: 1 2
  48. 48
    TAFINLAR (dabrafenib) capsules / tablets for oral suspension: Prescribing InformationNovartis Pharmaceuticals Corporation · DailyMed, U.S. National Library of Medicine · 2026
    Source Back to text
  49. 49
    Tumour-agnostic efficacy and safety of selpercatinib in patients with RET fusion-positive solid tumours other than lung or thyroid tumours (LIBRETTO-001): a phase 1/2, open-label, basket trialSubbiah V, Wolf J, Konda B, Kang H, Spira A, Weiss J, et al · Lancet Oncol · 2022;23:1261-1273
    DOIPubMed Back to text
  50. 50
    RETEVMO (selpercatinib) capsules / tablets: Prescribing InformationEli Lilly and Company · DailyMed, U.S. National Library of Medicine · 2026
    Source Back to text
  51. 51
    ENHERTU (fam-trastuzumab deruxtecan-nxki) for injection, for intravenous use: Prescribing InformationDaiichi Sankyo Inc · DailyMed, U.S. National Library of Medicine · 2026
    Source Back to text

This page is based on reputable scientific sources and was last updated on 28 October 2026. The content of this website is for educational and informational purposes only and does not replace consultation, diagnosis or treatment by a specialist physician. Consult your treating physician about any treatment decision.

Printed from the website; the full and up-to-date version: https://drdabiri.com/en/learn/reading-trials-biomarkers/

Related pages

26 October 2026 10 cheap steps that save patients' lives Ten cheap, accessible actions with strong evidence for emergency, oncology, surgical and primary care doctors: from immediate tretinoin in APL and antibiotics within 1 hour to zoledronic acid, BRCA, EGFR and MMR testing, morphine and short-course radiotherapy. 27 October 2026 Management of brain metastases Guideline-based management of brain metastases: radiosurgery for 1 to 4 lesions, hippocampal-avoidance whole-brain radiotherapy with memantine, and systemic treatment in EGFR/ALK lung cancer and BRAF melanoma. 27 October 2026 Cancer of unknown primary (CUP) A guide to assessing and treating cancer of unknown primary: an immunohistochemistry panel, imaging, favourable subsets with specific treatment, and the place of molecular profiling.