According to findings published in Drugs – Real World Outcomes, investigators analyzed longitudinal data from eighty-two adult participants within the National Institutes of Health All of Us Research Program who initiated systemic therapies between 2017 and 2023. Patients treated with checkpoint inhibitor regimens exhibited a median time to treatment discontinuation of 8.0 months compared with 1.4 months among patients receiving regorafenib or trifluridine/tipiracil. However, this extended treatment duration was coupled with higher drug acquisition and total clinical management expenses, generating an incremental cost-effectiveness ratio of 518,000 dollars per quality-adjusted life year gained. This economic estimate sits well above conventional United States willingness-to-pay benchmarks, indicating that while immunotherapy offers meaningful real-world treatment durability, its present pricing structure presents substantial cost-effectiveness hurdles for public healthcare payers.
Colorectal cancer continues to represent a leading cause of cancer-related mortality across the United States, with a substantial proportion of patients eventually progressing to metastatic or unresectable disease following initial exposure to standard fluoropyrimidine-based first-line and second-line combination regimens. For individuals requiring third-line systemic interventions, established therapeutic alternatives such as the multikinase inhibitor regorafenib or the oral cytotoxic combination trifluridine/tipiracil historically confer only modest survival extensions, typically demonstrating median overall survival intervals between 6.4 and 7.1 months in pivotal clinical trials. Nevertheless, the clinical durability, associated healthcare utilization, and comparative economic consequences of deploying these immunological therapies across routine, demographically diverse clinical environments have remained insufficiently characterized outside tightly controlled clinical trials.
To investigate these questions within routine oncology practice, the study team utilized the Version 8 Curated Data Repository of the National Institutes of Health All of Us Research Program. As of February 2025, this national research resource had enrolled more than 800,000 participants, with approximately 45 percent identifying as members of racial or ethnic minority groups historically underrepresented in biomedical literature. Among enrolled participants, more than 440,000 maintained linked electronic health records and more than 568,000 contributed biospecimens across a provider network that includes regional academic medical centers, federally qualified community health centers, and Veterans Affairs medical facilities. While broad and demographically diverse, the repository is not population-weighted and therefore does not represent a statistically representative cross-section of the broader United States population. The research protocol extracted longitudinal diagnosis codes, medication administrations, procedural encounters, laboratory measurements, histological designations, and vital status details, harmonizing all electronic health record entries into the standardized Observational Medical Outcomes Partnership common data model.
The study population was restricted to adult patients aged eighteen years or older diagnosed with stage IV or metastatic colorectal cancer who initiated third-line systemic therapy between January 2017 and October 2023. To construct robust comparative cohorts, third-line treatment was pragmatically identified by documenting the initiation of either an immune checkpoint inhibitor or regorafenib/trifluridine following prior documented exposure to fluorouracil. Because fluorouracil-based doublets paired with oxaliplatin or irinotecan represent the universal cornerstone of initial systemic therapy in advanced colorectal cancer, documented fluorouracil exposure was mandated to ensure that subsequent index treatments represented third-line or later interventions. Patients were excluded if they had more than one documented primary malignancy, treatment records preceding their primary cancer diagnosis, contradictory diagnostic codes, or fewer than twenty-one days of continuous exposure to their index drug. This twenty-one-day threshold was selected to match typical clinical dosing cycles, such as the standard three-week on schedule for regorafenib and three-week dosing intervals for checkpoint inhibitors, thereby reducing distortion from early discontinuations caused by immediate intolerance or administrative misclassification.
The final analytic cohort comprised eighty-two eligible individuals, including fifty patients in the immune checkpoint inhibitor cohort and thirty-two patients in the regorafenib or trifluridine/tipiracil cohort. Patients within the immunotherapy group were treated with pembrolizumab or nivolumab, administered either as single-agent therapies or in combination with ipilimumab. Baseline demographic characteristics revealed that the median age was slightly higher among patients receiving immunotherapy compared with the oral therapy cohort, reaching statistical significance with a p-value of 0.038. Sex and racial distributions remained generally balanced between both treatment groups, although individual subcategory counts below twenty-five patients were formally masked in compliance with data privacy mandates. Marked differences emerged in prior systemic treatment histories: patients in the oral regorafenib and trifluridine cohort were significantly more likely to have documented prior exposure to oxaliplatin, irinotecan, bevacizumab, and anti-epidermal growth factor receptor agents, with all comparative p-values falling below 0.001. Furthermore, structured documentation of molecular testing was exceedingly scarce, and zero patients had recorded microsatellite instability test results in standardized electronic tables.
The primary clinical effectiveness endpoint evaluated by the researchers was time to treatment discontinuation, defined as the duration in days from index therapy initiation to either the last documented drug administration without subsequent records for sixty days or the recorded initiation of an alternative systemic regimen. Patients remaining on continuous therapy within sixty days of the administrative database cutoff on October 1, 2023, were formally censored at their last administration date. Over an overall median follow-up period of 4.1 months, the median time to treatment discontinuation was 8.0 months in the immunotherapy cohort versus 1.4 months in the regorafenib/trifluridine cohort. In an unadjusted Cox proportional hazards model, immune checkpoint inhibitor administration was associated with a hazard ratio for treatment discontinuation of 0.49, with a 95 percent confidence interval spanning 0.30 to 0.79 and a p-value of 0.0037. This represented an estimated 51 percent reduction in the relative risk of discontinuing therapy, although formal residual testing suggested a departure from proportional hazards assumptions that warrants cautious, descriptive interpretation.
Resource utilization patterns were evaluated at the individual patient-event level across predefined clinical encounter categories, restricting calculations to individuals with documented encounters to capture true resource use. Outpatient visits, emergency room encounters, and inpatient hospital admissions were numerically higher among patients undergoing immune checkpoint inhibitor therapy than among those receiving oral agents, whereas the average inpatient length of stay showed no significant difference between the two therapeutic cohorts. Non-drug healthcare services, encompassing diagnostic imaging such as computerized tomography and magnetic resonance imaging, laboratory assessments including carcinoembryonic antigen testing, colonoscopies, and unscheduled hospital visits, produced cumulative utilization costs that were statistically comparable between treatment cohorts.
Economic outcomes were evaluated from the formal perspective of the United States Medicare payer across an analytic time horizon strictly restricted to observed treatment duration without discounting. Pharmaceutical acquisition expenses were calculated based on 2023 Centers for Medicare and Medicaid Services fee schedules, applying Part B Average Sales Price payment limits incorporating statutory six-percent reimbursement for infused biologics and Part D national average spending data for self-administered oral tablets. Medication costs were applied on a per-cycle basis and proportionally adjusted for fractional treatment cycles according to observed treatment persistence, without incorporating assumptions regarding dose reductions, confidential manufacturer rebates, or vial wastage. Non-drug healthcare encounters were priced using 2023 national reimbursement averages derived from the Medicare Physician Fee Schedule, the Inpatient Prospective Payment System, and the Clinical Laboratory Fee Schedule, ensuring a standardized, transparent costing methodology across all participating clinical centers.
Reflecting the substantial difference in therapy persistence and drug pricing, mean drug acquisition expenditures reached 207,558 dollars for the immune checkpoint inhibitor cohort compared with 119,805 dollars for the regorafenib and trifluridine cohort, a difference that was statistically significant with a p-value of 0.001. Mean total healthcare expenditures, combining drug acquisition and medical utilization, were 476,729 dollars for patients receiving immunotherapy versus 264,225 dollars for those prescribed oral therapies, yielding a p-value of 0.018. Quality-adjusted life years were derived by multiplying each individual participant’s duration of treatment by a validated colorectal cancer health-state utility weight of 0.75, which yielded 0.84 quality-adjusted life years for the immunotherapy arm versus 0.43 for the oral comparator arm. Dividing the incremental cost difference by the incremental gain in quality-adjusted survival produced a base-case incremental cost-effectiveness ratio of 518,000 dollars per quality-adjusted life year gained for third-line checkpoint inhibitors.
To rigorously characterize decision uncertainty surrounding the economic findings, the researchers implemented nonparametric bootstrap resampling using 1,000 stratified replicates that maintained the original cohort sample sizes. Across cost-effectiveness acceptability curves, immune checkpoint inhibitors demonstrated an estimated probability of cost effectiveness approaching zero percent at conventional United States willingness-to-pay thresholds between 50,000 and 150,000 dollars per quality-adjusted life year. The calculated probability of cost effectiveness rose to approximately 50 percent at an elevated willingness-to-pay threshold of 500,000 dollars per quality-adjusted life year, and reached roughly 80 percent only when willingness-to-pay values met or exceeded 1,000,000 dollars. Consistent with these probabilistic estimates, incremental net monetary benefit curves remained negative across all threshold values below 500,000 dollars. One-way sensitivity analyses varying the applied health-state utility weight across a range from 0.6 to 1.0 confirmed that while higher utility weights decreased the nominal cost per quality-adjusted life year, the incremental ratio consistently remained well above standard economic benchmarks.
The authors highlighted several critical methodological constraints that contextualize these real-world findings and mandate that the results be viewed as exploratory rather than confirmatory. The total analytic cohort was modest in size with eighty-two patients, precluding extensive multivariable adjustment and leaving findings vulnerable to unmeasured confounding, particularly regarding baseline functional performance status and precise anatomic stage at third-line initiation, which are not systematically captured in electronic health record tables. In addition, marked imbalances in documented prior chemotherapy regimens between cohorts likely reflect real-world treatment selection patterns rather than randomized clinical equivalence. Crucially, molecular biomarker documentation in structured electronic tables was incomplete, with microsatellite instability testing unrecorded across all participants due to clinical reliance on narrative pathology notes and unstructured image files. As the NIH All of Us data repository matures and refines structured molecular reporting, future investigations will be better equipped to evaluate the long-term clinical and economic value of checkpoint inhibitors within strictly biomarker-defined patient subgroups.
Subject of Research: Real-world treatment duration and cost-effectiveness of third-line therapies in advanced colorectal cancer
Article Title: Time to Treatment Discontinuation and Cost Effectiveness of Third-Line Therapies in Advanced Colorectal Cancer: Real-World Evidence from the NIH All of Us Research Program
Article References: Kiel, P. J., McGiffin, M. W., Skaar, T. C., & Preston, M. A. (2026). Time to Treatment Discontinuation and Cost Effectiveness of Third-Line Therapies in Advanced Colorectal Cancer: Real-World Evidence from the NIH All of Us Research Program. Drugs – Real World Outcomes, 13(1), 121-131. https://doi.org/10.1007/s40801-026-00539-9
Image Credits: AI Generated
DOI: 10.1007/s40801-026-00539-9
Keywords: colorectal cancer, immune checkpoint inhibitors, regorafenib, trifluridine-tipiracil, time to treatment discontinuation, cost-effectiveness analysis, NIH All of Us Research Program, incremental cost-effectiveness ratio, Medicare, quality-adjusted life years, pharmacoeconomics, electronic health records, real-world evidence, pembrolizumab, nivolumab, third-line systemic therapy, biomarker documentation
News Source: Nathaniel Bowman. (November 12, 2026). Third-Line Immunotherapy Extended Treatment Persistence but Exceeded Conventional Cost-Effectiveness Thresholds. Scienmag.



