Genomic Therapy Matching in Rare and Refractory Cancers
Frank P. Lin, Subotheni Thavaneswaran, John P. Grady, Christine E. Napier, Maya Kansara, Lucille Sebastian, Damien Kee, Jayesh Desai, Milita Zaheed, Sarah Chinchen, Samantha R. Oakes, James Blackburn, Hamish S. Scott, Anthony Glover, Stephen B. Fox, David Goldstein, Paul Leo
TL;DR
This study finds that matching cancer treatments to genomic biomarkers improves survival only when supported by strong clinical trial evidence.
Contribution
The study introduces a tiered framework for genomic therapy matching and shows its survival benefits in advanced cancers.
Findings
Patients with biomarker-matched therapies based on prospective trials had 21.2 months median survival vs 12.8 months for unmatched therapies.
Therapies matched using preclinical or repurposed evidence did not show survival benefits.
Repurposed therapies without direct evidence were associated with shorter survival.
Abstract
This cohort study assesses whether a tiered, evidence-based framework for matching genomic biomarkers to therapies is associated with differential overall survival in patients with advanced solid tumors. In patients with advanced, refractory cancers, is a tiered framework for matching genomic biomarkers to therapies associated with improved survival? In this cohort study of 3383 patients with advanced solid tumors, those with biomarkers supported by prospective trial evidence who received matched therapy had a median overall survival of 21.2 months vs 12.8 months for those receiving unmatched therapy. No survival benefit was observed for therapies matched using preclinical evidence or repurposed from other cancer types. These findings support prioritizing genomically guided therapies based on the strength of clinical evidence, as only matches supported by prospective trials were…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4| Characteristic | No. (%) | ||
|---|---|---|---|
| Cohort A (n = 2065) | Cohort B (n = 1318) | Total (N = 3383) | |
| Age group, y | |||
| <20 | 8 (0.4) | 10 (0.8) | 18 (0.5) |
| 20-34 | 148 (7.2) | 119 (9.0) | 267 (7.9) |
| 35-44 | 231 (11.2) | 171 (13.0) | 402 (11.9) |
| 45-54 | 383 (18.5) | 262 (19.9) | 645 (19.0) |
| 55-64 | 570 (27.6) | 366 (27.8) | 936 (27.7) |
| 65-74 | 529 (25.6) | 310 (23.5) | 839 (24.8) |
| 75-84 | 186 (9.0) | 79 (6.0) | 265 (7.8) |
| ≥85 | 10 (0.5) | 1 (0.1) | 11 (0.3) |
| Sex | |||
| Female | 1072 (51.9) | 720 (54.6) | 1792 (53.0) |
| Male | 993 (48.1) | 598 (45.4) | 1591 (47.0) |
| ECOG group | |||
| 0 | 884 (42.8) | 706 (53.6) | 1590 (47.0) |
| 1 | 1072 (51.9) | 575 (43.6) | 1647 (48.7) |
| 2 | 94 (4.6) | 31 (2.4) | 125 (3.7) |
| 3 | 1 (<0.1) | 1 (0.1) | 2 (<0.1) |
| Not reported | 1 (<0.1) | 0 | 1 (<0.1) |
| Urban-rural classification | |||
| Urban | 1852 (89.7) | 1214 (92.1) | 3066 (90.6) |
| Rural | 212 (10.3) | 102 (7.7) | 314 (9.3) |
| Other | 1 (<0.1) | 2 (<0.1) | 3 (<0.1) |
| Rare class | |||
| Common | 442 (21.4) | 285 (21.6) | 727 (21.5) |
| Less common | 167 (8.1) | 62 (4.7) | 229 (6.8) |
| Rare | 1456 (70.5) | 971 (73.7) | 2427 (71.7) |
| Cancer type | |||
| Sarcoma | 354 (17.1) | 300 (22.8) | 654 (19.3) |
| Colorectal | 205 (9.9) | 141 (10.7) | 346 (10.2) |
| Pancreas | 185 (9.0) | 103 (7.8) | 288 (8.5) |
| High-grade gliomas | 146 (7.1) | 93 (7.1) | 239 (7.1) |
| Breast | 104 (5.0) | 85 (6.4) | 189 (5.6) |
| Ovarian | 90 (4.4) | 82 (6.2) | 172 (5.1) |
| Biliary and gallbladder | 113 (5.5) | 57 (4.3) | 170 (5.1) |
| Gastroesophageal | 105 (5.1) | 44 (3.3) | 149 (4.4) |
| Unknown primary | 79 (3.8) | 54 (4.1) | 133 (3.9) |
| Non–small cell lung | 78 (3.8) | 36 (2.7) | 114 (3.4) |
| Endometrial | 64 (3.1) | 33 (2.5) | 97 (2.9) |
| Prostate | 53 (2.6) | 40 (3.0) | 93 (2.7) |
| Bladder | 44 (2.1) | 32 (2.4) | 76 (2.2) |
| Head and neck | 53 (2.6) | 13 (1.0) | 66 (2.0) |
| Small bowel and appendix | 42 (2.0) | 20 (1.5) | 62 (1.8) |
| Salivary gland | 40 (1.9) | 21 (1.6) | 61 (1.8) |
| Kidney | 31 (1.5) | 15 (1.1) | 46 (1.4) |
| Other | 279 (13.5) | 149 (11.3) | 428 (12.7) |
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsCancer Genomics and Diagnostics · Lung Cancer Treatments and Mutations · Colorectal Cancer Treatments and Studies
Introduction
Genomic profiling has fundamentally altered the therapeutic landscape in oncology, propelling biomarker-centric drug development while leading to the regulatory approval of tumor-agnostic therapies.^1,2,3,4,5,6,7,8^ Clinically, large multigene panel testing is increasingly used to identify treatment options for patients with advanced cancers who have exhausted standard therapies. Although studies using genomic biomarker selection have reported improved disease control,^9,10,11,12,13,14,15,16^ clinical outcomes remain inconsistent across different programs.^17,18,19,20,21^ A primary contributor to this variability is the lack of a standardized definition of actionability of genomic alterations. Methodologies for determining actionability vary widely and include the consensus of molecular tumor boards,^16,18,22,23,24,25^ algorithmic matching scores,^26^ and pathway-based analyses.^16,22^ Heterogeneous interpretation can lead to inconsistent recommendations, complicating the evaluation of clinical utility of genomic biomarker testing.^24^
We report the outcomes from the screening component of the national Molecular Screening and Therapeutic (MoST) program in Australia.^25^ The primary objective of the MoST program is to use genomic profiling of archival tissue to identify actionable biomarkers for patients with advanced, rare, or neglected cancers. To address the challenge of inconsistent biomarker interpretation, we previously developed the Therapy-Oriented Precision Oncology Guidelines for Recommending Anticancer Pharmaceuticals (TOPOGRAPH) framework.^27^ This framework is based on a curated catalog of biomarker-linked therapies, with each therapy categorized by the strength of its clinical evidence and its stage of drug development.
The primary aim of this cohort study is to report the outcomes of the MoST screening program. A key secondary aim is to determine whether the TOPOGRAPH framework can stratify clinical outcomes for patients with cancer receiving genomically matched therapy, thereby informing the development of a robust system to guide therapy selection for this population with limited treatment options.
Methods
Study Type and Oversight
This cohort study was conducted within the MoST program, a national precision oncology initiative,^25^ approved by the St Vincent’s Hospital Human Research Ethics Committee. All participants provided written informed consent prior to enrollment. The study is registered with the Australian New Zealand Clinical Trials Registry (ACTRN12616000908437) and followed Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
Study Population and Eligibility
Eligible participants were adults (≥18 years old) with a pathologically confirmed metastatic or advanced solid tumor for which standard systemic therapies had been exhausted or were not tolerated. Inclusion required an Eastern Cooperative Oncology Group (ECOG) Performance Status of 0 to 2. Key exclusion criteria were uncontrolled comorbidities or symptomatic central nervous system involvement. Patients with a history of another cancer within 2 years prior to enrollment were excluded, unless the prior cancer had been definitively treated and was progression free for 6 months.
Study Cohort
Consecutive patients who underwent genomic profiling between June 2016 and December 2021 were screened. This study period was selected to align with the nationwide implementation of a cancer genomics program that stipulated a recruitment target of 3095 patients. Patients were excluded from the analysis for poor tissue quality, sequencing failure, or death prior to the availability of genomic profiling results. Clinical and demographic data, primary diagnosis, ECOG Performance Status, and Charlson Comorbidity Index were collected at the time of consent. The number of prior lines of therapy, defined by unique systemic treatment or combinations, was determined by 3 investigators (F.P.L., S.T., and C.E.N.).
The study population was stratified into 2 cohorts based on whether subsequent systemic therapy was administered after genomic results were returned: cohort A received no further therapy, and cohort B received 1 or more lines of therapy following genomic profiling. This stratification was carried out to facilitate a meaningful analysis of treatment outcomes, as considerable heterogeneity in survival characteristics was observed across the screened population (eMethods in Supplement 1).
Genomic Profiling, Tiering, and Therapy Matching Strategy
Archival formalin-fixed, paraffin-embedded diagnostic material was sequenced using next-generation sequencing panels, primarily the Illumina TruSight Oncology (TST170 and TSO500) and the FoundationOne CDx panels. Genomic variants were reviewed at weekly molecular tumor board meetings (fortnightly at Western Australian sites).
The level of evidence for therapy-biomarker associations was classified using the TOPOGRAPH knowledge base (version AU 20220828).^27^ Systemic therapies matched to genomic variants were stratified into hierarchical tiers: tiers 1 to 2 (approved for use with the specific biomarker in the cancer type), tier 3 (proven clinical activity in the cancer type but not yet standard care), tier 3B (therapies with established efficacy for the same biomarker in a different cancer type but lacking direct evidence in the specific histotype), tier 4 (preclinical or early clinical evidence), and tier R2 (therapies predicted to be inactive).
For patients in cohort B, the overall tier of the therapy received after genomic profiling was determined according to a predefined hierarchy for identifying the most active therapy: (1) the therapy with the highest TOPOGRAPH tier among those received subsequent to genomic profiling; (2) for multiple therapies genomically matched to the same tier, the earliest therapy administered; or (3) classifying the therapy as unmatched if it had no corresponding genomic biomarker. The tiering process was performed algorithmically to prevent subjective bias in tier assignment.^28^
Outcome Measures
The primary outcome was overall survival (OS), defined as the time from receipt of molecular profiling results to death from any cause. Participants underwent follow-up assessments at 6 and 12 months after sequencing, then annually for up to 5 years. For the OS analysis, data were censored at July 20, 2022. Patients of unknown survival status were censored at their last follow-up date. Secondary outcomes included the frequency and type of actionable molecular alterations, the proportion of patients who received genomically matched therapy, and the molecular profiling turnaround time (days from sample receipt to report finalization).
Statistical Analysis
Baseline characteristics were summarized using descriptive statistics: medians with IQRs for continuous variables and frequencies for categorical variables. Differences between patient groups were assessed using the Mann-Whitney U test (continuous variables) and the χ^2^ test (categorical variables).
The primary analysis compared OS between patients who received genomically matched therapy and those who received nonmatched therapy. Matched therapies were categorized into evidence tiers 1 to 4 based on the level of evidence defined in the TOPOGRAPH database. OS was estimated using the Kaplan-Meier method; between-group difference was compared using the log-rank test. Median follow-up time was calculated with the reverse Kaplan-Meier method. Prespecified subgroup analyses of OS were conducted to compare (1) patients receiving any tier-matched therapy (tiers 1-4) vs those with an actionable alteration who received only nonmatched therapy, (2) patients receiving therapies with established clinical activity (tiers 1-3) vs those receiving investigational therapies (tiers 3B/4), and (3) patients within individual tier groups (tiers 1-2, 3, 3B, and 4).
To adjust for confounders and time-dependent biases related to the timing of therapy initiation, a multivariable Cox proportional hazards model was constructed, incorporating a time-dependent covariate for the initiation of the most active matched therapy. The following baseline prognostic factors were also adjusted: age, ECOG Performance Status, cancer type, prior lines of therapy, and prior receipt of a tier-matched therapy. The primary result of this model was the adjusted hazard ratio (aHR) for death, reported with 95% CIs.
R, version 4.0.5 (R Project for Statistical Computing), was used for descriptive, survival, and regression analyses. All statistical tests were 2-sided, and P < .05 was considered statistically significant. Data were analyzed from July 2022 to July 2024.
Results
Study Population
The MoST program recruited 4173 heavily pretreated patients with rare and refractory advanced solid tumors between June 2016 and December 2021 (Figure 1). After exclusions, results from 3383 patients (81.1%; mean [SD] age, 57.1 [14.3] years; 1792 [53.0%] female) were eligible to be included for analysis at data cutoff. The most frequent cancers were sarcoma in 654 patients (19.3%), colorectal cancer in 346 (10.2%), pancreatic cancer in 288 (8.5%), high-grade gliomas in 239 (7.1%), and breast cancer in 189 (5.6%). Overall, 2427 patients (71.7%) had rare cancers (<6 per 100 000 population frequency), and 229 (6.8%) had less common cancers (6-12 per 100 000 population frequency) (Table and eTable 1 in Supplement 1).
Cohort Flow Diagram and Genomic Matching Tier DefinitionsPrognostic evaluations were conducted for both cohorts A and B. In cohort B, tier-matched analyses were also conducted to assess the potential treatment effect of genomically matched therapies, stratified by tiers. Further exploratory analyses were conducted to identify potential genomic alterations that may be predictive of differential treatment effects. Therapies, together with their genomic profiling results and cancer type, were searched against the TOPOGRAPH (Therapy-Oriented Precision Oncology Guidelines for Recommending Anticancer Pharmaceuticals) knowledge base. This knowledge base comprises triples of biomarker variants, cancer type, and drug, all linked by supporting evidence. A therapy is considered matched if any drug from its regimen class is present in the TOPOGRAPH curation, mirroring the clinical strategy used in therapy selection based on genomic biomarker testing. CGP indicates comprehensive genomic profiling; FDA, US Food and Drug Administration; MoST, Molecular Screening and Therapeutic program; MTB, molecular tumor board; PBS, Pharmaceutical Benefits Scheme; TGA, Therapeutic Goods Administration.
Participant Characteristics
Eligible patients were stratified into 2 cohorts (Figure 1): 2065 patients (61.0%) who received no further systemic therapies (cohort A) and 1318 patients (39.0%) who received systemic therapy after genomic profiling (cohort B). Compared to cohort B, patients in cohort A were older (mean [SD] age, 55.7 [14.3] vs 58.0 [14.2] years; P < .001), had worse performance status (ECOG group 0: cohort A, 884 of 2065 [42.8%]; cohort B, 706 of 1318 [53.6%]; P < .001), and had greater comorbidities (mean [SD] Charlson Comorbidity Index, 1.38 [1.52] vs 1.63 [1.67]; P < .001) at enrollment. There was not a statistically significant difference in the mean (SD) number of lines of systemic therapy received prior to genomic profiling between cohorts A and B (1.71 [1.52] vs 1.70 [1.49]; P = .73).
The patients were followed for a median (IQR) of 22.6 (11.1–35.4) months, with 2131 deaths (63.0%). The median OS after genomic profiling was 11.3 months (95% CI, 10.5-11.9 months). A shorter median OS was observed in cohort A (8.2 months; 95% CI, 7.4-9.0 months) compared to cohort B (14.1 months; 95% CI, 13.4-15.2 months). In cohort B, the median time to first systemic therapy was 2.4 months (95% CI, 2.2-2.6 months). The median (IQR) time from study consent to the return of genomic profiling results was 49 (36-64) days (mean, 71.5 days). The median OS from study consent was 13.3 months (95% CI, 12.7-14.1 months): 10.4 months (95% CI, 9.6-11.5 months) in cohort A and 16.8 months (95% CI, 15.6-17.8 months) in cohort B. The landscape of alterations among the entire cohort is shown in eFigure 1 in Supplement 1.
Therapy-Associated TOPOGRAPH Tiers Based on Genomic Results
There were 3003 patients (88.8%) with 1 or more genomic biomarkers for which a tier (1-4) matched therapy could be assigned (tier-assigned group). Clinically active biomarker-linked therapies (tier 1-3A) were identified in 1270 patients (37.5%). The proportion of patients with no tiered genomic biomarker (untiered group) was higher in cohort A than in cohort B (257 [12.5%] vs 123 [9.3%]; P = .006). In the tier-assigned group, no difference was noted in the distribution of tiers between cohorts A and B (eTable 2 in Supplement 1). For the clinically active tier group in cohort B (tiers 1-3A; 526 of 3383 patients [15.5%]), the median (IQR) of matched biomarker-linked drug classes per patient was 0 (0-1) (mean, 1.0) but with a maximum of 22 actionable drug classes. The median (IQR) number of actionable drug classes matched to the investigational therapy tier group (Tier 3B and 4; 669 of 3383 patients [19.8%]) per patient was 7 (3-15) (mean, 10.1; maximum, 52).
Survival Association Between Harborage of Actionable Biomarkers
For patients who did not undergo further therapy (cohort A), carriage of tier 1 to 4 genomic alteration was associated with a statistically significant shorter OS compared to patients without a matching tier (median, 7.1 months [95% CI, 6.5-7.7 months] vs 30.1 months [95% CI, 16.6 months to not reached]; unadjusted HR, 2.39; 95% CI, 1.93-2.97; P < .001; Figure 2A). Similar results were obtained for patients in cohort B who received unmatched therapies, with those in tier 1 to 4 having a median OS of 13.7 months (95% CI, 12.8-14.6 months) vs 20.3 months (95% CI, 16.2-24.2 months) for the untiered group (unadjusted HR, 1.43; 95% CI, 1.13-1.82; P < .001; Figure 2B). No differences in survival were observed when comparing individual tiers (Figure 2C and D). The differences in OS between tiered and untiered groups remained statistically significant in cohort A after adjusting for stratified age group, ECOG Performance Status, cancer type, number of lines of prior therapy, and Charlson Comorbidity Index (aHR, 1.75; 95% CI, 1.38-2.21; P < .001), but not in cohort B (aHR, 1.21; 95% CI, 0.95-1.55; P = .12) (eFigure 2 in Supplement 1). Cohorts A and B showed a similar distribution of cancer types by frequency and number of prior lines of therapy (eFigures 3 and 4 in Supplement 1).
Kaplan-Meier Curves Among Distinct Prognostic Groups Stratified by Matching Status in Patients With Advanced Cancers Undergoing Genomic ProfilingThis figure presents the analysis of overall survival in cohorts A and B from the time of genomic profiling, by whether the genomic profiles were matched to any TOPOGRAPH (Therapy-Oriented Precision Oncology Guidelines for Recommending Anticancer Pharmaceuticals) tier or stratified. Overall survival is measured from the date of genomic profiling. The forest plot showing adjusted hazard ratios of prognostic groups is shown in eFigure 2 in Supplement 1.
Survival Outcomes in Patients Who Underwent Further Therapy After Profiling
Among the 1318 patients in cohort B, 382 (29.0%) received 1 or more lines of systemic therapy after genomic profiling, matched to tier 1 to 4 biomarkers, with the remainder receiving unmatched therapies. Compared to 936 patients receiving only unmatched therapy regardless of biomarker status, the OS was longer for patients who received matched therapies (median, 13.4 months [95% CI, 12.7-14.7 months] vs 16.3 months [95% CI, 13.8-18.4 months]; aHR, 0.76; 95% CI, 0.65-0.88; P < .001; eFigure 5 in Supplement 1). In 116 patients (8.9%) who received 1 or more clinically active matched therapies (tiers 1-3A), a statistically significant longer OS was observed (median, 21.2 months [95% CI, 17.1-26.8 months]; aHR, 0.54; 95% CI, 0.41-0.71, P < .001; eFigure 5 in Supplement 1). In contrast, 266 patients who received an investigational-matched therapy (tier 3B and 4) had a median OS of 14.2 months (95% CI, 12.7-16.8 months; aHR, 0.86; 95% CI, 0.72-1.02; P = .09). In 21 patients who received a therapy classified as clinically inactive (tier R2), the median OS was 9.7 months (95% CI, 5.3-17.1 months; aHR, 0.89; 95% CI, 0.50-1.60; P = .70).
Estimating Treatment Effect of Matched Therapy Within Tier Groups in Cohort B
For those with tier 1 to 4 genomic profiles, 382 patients who received matched therapies had a statistically significant longer OS over the 813 patients who received unmatched therapy (median, 16.3 months [95% CI, 13.8-18.4 months] vs 12.8 months [95% CI, 12.1-13.8 months]; aHR, 0.78; 95% CI, 0.66-0.91; P = .002; Figure 3A). In the group of patients carrying a biomarker associated with at least 1 tier 1 to 3(A) therapy, the 116 patients who received a matched therapy had statistically significant longer OS than the 410 patients who received unmatched therapy (21.2 months [95% CI, 17.1-26.8 months] vs 12.8 months [95% CI, 11.7-13.9 months]; aHR, 0.60; 95% CI, 0.44-0.82; P = .001; Figure 3B). The magnitude of improvement in OS was comparable for receipt of matched vs unmatched therapies between tiers 1 to 2 (drugs approved by regulatory agencies) and tier 3 findings (eFigure 6 in Supplement 1). Among the 64 patients who received tier 1 to 2 matched therapies, median OS was 22.4 months (95% CI, 16.1-26.8 months) vs 13.2 months (95% CI, 10.9-15.3 months) for the 172 patients who received therapies not matched to tiers 1 to 2 (aHR 0.74; 95% CI, 0.48-1.14; P = .17). In the tier 3 group, matched therapy was associated with a median OS of 18.8 months (95% CI, 12.5-40.3 months) vs 12.8 months (95% CI, 11.5-14.6 months) among patients who received unmatched therapies (aHR, 0.53; 95% CI, 0.31-0.90; P = .02; Figure 3C).
Kaplan-Meier Curves Among the Tier-Matched Analysis in Cohort B Comparing Survival OutcomesA, Difference in overall survival between the matched (tiers 1-4) and unmatched therapy groups (adjusted hazard ratio, 0.76; 95% CI, 0.70-0.90; P < .001). B, Analysis of patients whose highest genomically matched TOPOGRAPH (Therapy-Oriented Precision Oncology Guidelines for Recommending Anticancer Pharmaceuticals) tier was in the clinically active group (tiers 1-3) (adjusted hazard ratio, 0.60; 95% CI, 0.50-0.80; P < .001). C, Analysis of the investigational therapy tier group (tiers 3B/ 4) (adjusted hazard ratio, 0.83; 95% CI, 0.70-1.00; P = .10).
For patients with investigational tier biomarkers (tiers 3B and 4), there was no statistically significant difference in OS among the 133 patients who received matched treatments compared to the group who received only unmatched therapy after adjusting time to therapy in a time-dependent Cox regression model (median, 14.5 months [95% CI, 12.6-18.4 months] vs 12.8 months [95% CI, 12.0-14.7 months]; aHR, 1.04; 95% CI, 0.84-1.29; P = .71; Figure 4). Notably, there was not a statistically significant longer OS among patients who received matching therapies in tier 3B, compared with those who received unmatched therapy, based on genomic variants alone, with supporting evidence obtained only in noncognate histotypes (13.6 months [95% CI, 8.0-16.8 months] vs 12.5 months [95% CI, 11.3-13.5 months]; aHR 1.40; 95% CI, 1.00-1.96; P = .047; eFigure 6 in Supplement 1). Notably, there was statistically significant longer delay from genomic profiling to matched tier 3B therapy among patients who received matched therapy compared to those receiving unmatched therapy (median [IQR], 167 [59-310] days vs 75 [27-174] days; P = .01; eTable 3 in Supplement 1). For tier 4 (preclinical or early clinical evidence), median OS was 21.8 months (95% CI, 14.5-36.7 months) vs 16.2 months (95% CI, 12.8-19.1 months) (aHR, 0.77; 95% CI, 0.47-1.26; P = .30; eFigure 6 in Supplement 1). A consistent, tier-based OS difference was observed in (1) an exploratory tier-matched analysis selecting only the same therapies in both matched and unmatched groups (eFigure 7 in Supplement 1), (2) sensitivity analyses accounting for patients who died before receiving profiling results (eFigures 8 and 9 in Supplement 1), and (3) analyses using updated TOPOGRAPH tiering knowledge base (eFigure 10 in Supplement 1).
Forest Plots Showing the Unadjusted and Adjusted Hazard Ratios (HRs) from the Tier-Matched Analysis in Cohort B, Comparing Time-to-Matched Therapy With Time-to-Unmatched Therapy in Regression ModelsHRs were adjusted for the time to initiation of the most active subsequent therapies, age, Eastern Cooperative Oncology Group Performance Status at consent, cancer type, prior lines of therapy, and whether the patient had previously received a therapy that matched the corresponding TOPOGRAPH (Therapy-Oriented Precision Oncology Guidelines for Recommending Anticancer Pharmaceuticals) tier or tier group.
Discussion
A clear definition of therapy matching can guide oncologists in selecting among multiple biomarker-dependent treatments for patients with advanced cancers. Through the MoST program, we observed tier-dependent differences in both natural history and potential treatment effect based on genomic profiling. Allowing for the observational nature of this study, these results support implementing a strength-of-recommendation tiering framework as a strategy for rationally prioritizing additional lines of genomically informed therapy in heavily treated patients with solid malignant tumors, providing a pragmatic strategy for structured decision-making.
We observed a 40% reduction in the risk of death associated with the receipt of genomic matched therapy in the clinically active tier group (tiers 1-3A) after accounting for confounders and time-related biases. Assignment to this group was based on supportive evidence from diverse, prospective biomarker-linked trials with at least evidence from phase 2 (or equivalent) trials.^27^ Conversely, we found no OS difference in patients who received treatment matched to the investigational therapy tier groups overall (tiers 3B and 4). The observation of lack of survival difference regarding drug repurposing (tier 3B) was particularly noteworthy, as inferring treatments based on biomarkers alone from evidence extrapolated from a noncognate cancer type is commonly used in the absence of histotype-specific trials or regulatory approvals. Off-label repurposing of drugs should be discouraged outside of clinical trials, or at least prompt a discussion with the patient about options not informed by genomics to avoid ineffective treatment. The present results corroborate international guidelines.^29,30^
Through the MoST program, 37.5% of patients profiled carried a tier 1 to 3(A) biomarker, although only a minority (3.4%) received a matched therapy. While clinical factors could affect uptake of a therapy, part of this disparity could be attributed to drug access patterns in Australia, warranting services research to identify barriers that preclude treatment access. Notably, the magnitude of OS benefit for unapproved drugs in tier 3 was similar to approved therapies (tiers 1-2), which could advocate for expanded eligibility to access relevant biomarker-dependent trials to allow better equitable access. Earlier genomic profiling in the cancer journey may also allow patients timely access to appropriate investigational matched therapies.
For tier 4 matched therapies, the observed magnitude of OS differences corroborated the trend where modern phase 1 trials have seen higher response rates in recent years.^31^ Overall, this study reinforced the importance of ongoing trials and innovative precision medicine platforms to efficiently support or discontinue biomarker-drug pairs to refine the continuously changing evidence landscape in precision oncology, as recently illustrated in the Molecular Analysis for Therapy Choice (MATCH) trial.^32^ Continuing investment in platform trials complements randomized trials in rare cancers (70% of the MoST cohort), which account for more than 40% of cancer-related deaths in Australia.^33^
Limitations
This study has highlighted the value of a tier-based matching framework, although comparison with other level-of-evidence systems, including the European Society for Medical Oncology Scale for Clinical Actionability of Molecular Targets,^34^ would be necessary to understand the optimal matching strategy in future studies. This study also did not include patients who died before their genomic results were available, although a sensitivity analysis showed the magnitude of OS difference remained consistent with the primary analysis. Further examining tiering strategies should also incorporate nuanced interactions between evolving standards of care, genomic variants, tumor histology, and treatment histories to support standardized decision-making for prioritizing therapies and clinical trials.
Conclusions
In this cohort study of patients with solid tumors, analysis of the MoST screening program demonstrates that genomic matching is associated with observable differences in overall survival when underpinned by prospective clinical trial evidence. These findings also support the use of evidence-based frameworks for prioritizing therapeutic stratification in advanced cancers. Overcoming barriers to drug access and elucidating the interplay between molecular and histotypic response determinants, particularly in rare cancers, necessitates international collaboration. By pooling data to power expanded cohort studies, the global oncology community can refine recommendation standards and maximize the clinical utility of comprehensive genomic profiling.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Marabelle A, Fakih M, Lopez J, . 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 study. Lancet Oncol. 2020;21(10):1353-1365. doi:10.1016/S 1470-2045(20)30445-9 32919526 · doi ↗ · pubmed ↗
- 2Marabelle A, Le DT, Ascierto PA, . Efficacy of pembrolizumab in patients with noncolorectal high microsatellite instability/mismatch repair-deficient cancer: results from the phase II KEYNOTE-158 study. J Clin Oncol. 2020;38(1):1-10. doi:10.1200/JCO.19.02105 31682550 PMC 8184060 · doi ↗ · pubmed ↗
- 3Hong DS, Du Bois SG, Kummar S, . Larotrectinib in patients with TRK fusion-positive solid tumours: a pooled analysis of three phase 1/2 clinical trials. Lancet Oncol. 2020;21(4):531-540. doi:10.1016/S 1470-2045(19)30856-3 32105622 PMC 7497841 · doi ↗ · pubmed ↗
- 4Salama AKS, Li S, Macrae ER, . Dabrafenib and trametinib in patients with tumors with BRAFV 600E mutations: results of the NCI-MATCH trial subprotocol H. J Clin Oncol. 2020;38(33):3895-3904. doi:10.1200/JCO.20.00762 32758030 PMC 7676884 · doi ↗ · pubmed ↗
- 5Subbiah V, Wolf J, Konda B, . 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 trial. Lancet Oncol. 2022;23(10):1261-1273. doi:10.1016/S 1470-2045(22)00541-1 36108661 PMC 11702314 · doi ↗ · pubmed ↗
- 6Subbiah V, Lassen U, Élez E, . Dabrafenib plus trametinib in patients with BRAFV 600E-mutated biliary tract cancer (ROAR): a phase 2, open-label, single-arm, multicentre basket trial. Lancet Oncol. 2020;21(9):1234-1243. doi:10.1016/S 1470-2045(20)30321-1 32818466 · doi ↗ · pubmed ↗
- 7Subbiah V, Kreitman RJ, Wainberg ZA, . Dabrafenib plus trametinib in patients with BRAF V 600E-mutant anaplastic thyroid cancer: updated analysis from the phase II ROAR basket study. Ann Oncol. 2022;33(4):406-415. doi:10.1016/j.annonc.2021.12.014 35026411 PMC 9338780 · doi ↗ · pubmed ↗
- 8Wen PY, Stein A, van den Bent M, . Dabrafenib plus trametinib in patients with BRAFV 600E-mutant low-grade and high-grade glioma (ROAR): a multicentre, open-label, single-arm, phase 2, basket trial. Lancet Oncol. 2022;23(1):53-64. doi:10.1016/S 1470-2045(21)00578-7 34838156 · doi ↗ · pubmed ↗