Veristat
iostatisticians are misperceived as only appearing at the end of a clinical trial to analyze the collected data and present the results. Yet nothing could be further from the truth. In practice, these statistical analysts are active participants from the earliest stages of research planning until long after approval.
Biostatisticians play a particularly pivotal role in oncology trials, serving as an ever-present, important throughline in the complex and challenging research environment of oncology.
The Oncology Conundrum: Endpoints, Ethics, and Evidence
Oncology trials are typically complex and high stakes, with outcomes that can extend survival and enhance quality of life. Achieving a breakthrough therapy in oncology requires careful coordination of science, regulation, patient care, and data. Specifically, they pose a unique set of challenges for biostatisticians:
Endpoints are complex. Overall survival is the gold standard, but collecting survival data can become burdensome and incomplete. Expected survival time may be quite variable and subject to intervening alternative therapies and comorbidities as competing risks. Alternative endpoints such as progression-free survival, tumor shrinkage, or response rate are also commonly used in oncology research. These and similar endpoints aim to measure disease burden, though improvement may not directly map to a meaningful clinical benefit.
- Overall Survival (OS): the time from trial randomization until death from any cause.
- Expected Survival Time (EST): the estimated time an individual may live, typically estimated as the median of the survival distribution measured in a trial.
- Progression-Free Survival (PFS): the time from randomization until objective tumor progression or death, whichever occurs first.
- Response Rate (RR): the proportion of patients in a trial with tumor size reduction of a predefined amount that is sustained for a minimum period. RR can also be applied to biomarker measurements for some cancers.
Safety signals are noisy. Patients can experience abnormal lab values and serious health events from the cancer itself. When participating in clinical trials, patients may be treated with new, experimental therapies where researchers are still learning about the potential side effects of the products. It is critical to collect and evaluate all adverse reactions to drug products to ensure that new products brought to market are effective and safe. Distinguishing which adverse health events stem from the investigational therapy versus underlying disease is complex.
Given these challenges, oncology research demands both rigor and creativity. Often biostatisticians must innovate within critical constraints, developing credible ways to measure efficacy and safety when traditional approaches are insufficient.
Case Study: Designing an Ovarian Cancer Study
Ovarian cancer is a leading cause of death from gynecologic cancers worldwide, despite a high initial response rate to platinum and taxane treatment in patients with advanced cancer. Securing regulatory approval of ovarian cancer treatment can be challenging when it comes to creatively analyzing genetic subgroups, specifically patients with BRCA1/2 mutations and those with homologous recombination deficiencies.
In the pivotal trial for the PARP inhibitor niraparib, biostatisticians were instrumental in creating a sequential analysis strategy that demonstrated efficacy in the most responsive subgroup first, then in broader populations. This approach not only secured FDA approval but also resulted in a broader label for the drug than had been anticipated. Now, more women can access this critical therapy to extend remission of ovarian cancer by an average of 18 months.
Case Study: Proving Benefit in a Rare Cancer
Blastic plasmacytoid dendritic cell neoplasm (BPDCN) is a rare, aggressive hematologic cancer, a complex disease where efficacy is assessed via multipart response criteria, including measurements of bone marrow blasts, white blood cell and platelet counts, reduction in lymph node sizes, absence of palpable spleen or liver nodules, and clearance of skin lesions. To reach a complete response (CR), a patient must have resolution of abnormalities across all components. All other components may resolve to normal ranges ahead of skin lesion clearance, defined as a clinical complete response (CCR), a beneficial response to treatment.
FDA approved Elzonris for BPDCN in 2018 based on a nonrandomized, multistage, open-label, multicenter evaluation in 47 patients. The biostatisticians validated the primary outcome of CCR by using an innovative empirical evaluation method to demonstrate that response duration for patients who had CCR came from the same probability distribution as CR. This critical analysis strengthened the statistical power of the evidence base, supporting approval in both the US and Europe.
Adaptive Designs and Early Decisions
Biostatisticians have been at the forefront of adaptive and multicohort trial designs, which are particularly valuable in oncology research.
Adaptive trials allow for prospectively planned modifications to one or more aspects of the study design based on accumulating data from the trial itself. Adaptive designs can include modifications such as early stopping for efficacy or for futility, midstudy adjustments to sample size, endpoint selection, or the selection of treatment arms. In cancer research, adaptive designs with early stopping for futility can help to end trials earlier, providing an opportunity for patients to receive alternate treatments.
Additionally, adaptive designs with sample size re-estimation, population enrichment, and dropping of unpromising treatment arms increase the probability of a successful trial. Statisticians are critical to the development of an adaptive study design, the study objectives and endpoints, the clinical study protocol, and the statistical analysis plan and provide input including considerations around the sample size, sources of bias, and mitigation strategies so that the study will maintain its scientific integrity and provide reliable results without an inflated false positive rate.
Early-phase multicohort trials allow sponsors to evaluate multiple tumor types and/or dose levels in parallel, analyze interim results, and drop unsafe or ineffective cohorts while expanding promising ones, all under pre-specified statistical rules. These designs allow sponsors to prioritize cancer types for further development, establish initial evidence of efficacy earlier, and accelerate cancer development.
Both trial types can accelerate trials, conserve resources, and reduce patient exposure to ineffective treatments, while also meeting rigorous regulatory standards. Their success heavily depends on statistical thinking built into a trial from design, rather than applied after the fact.
Overall Survival versus Quality of Life
Balancing survival outcomes, tumor response measurements, and quality of life (QoL) ensures that a new treatment’s benefits outweigh the risks. Extending survival is not enough if patients are too sick to benefit from the time that they gain. Biostatisticians play an essential role in validating patient-reported outcomes by turning subjective experiences into measurable, reliable data. Without robust statistical frameworks, QoL measures risk being dismissed as anecdotal. With proper methodology, they carry weight alongside survival endpoints, ensuring therapies deliver both longer life and better living.
Biostatistician as Regulatory Advocate
Biostatisticians serve as frontline advocates in regulatory engagement. Their presence in regulatory meetings, in collaboration with clinical development scientists and medical experts, reassures agencies that trial design, analysis, and evidence are sound.
Biostatisticians can often anticipate questions from regulators about such things as dosing, endpoints, comparators, and potential bias. This helps regulators gain confidence in the research and strengthens the credibility of both the sponsor and the therapy under review. In many cases, the biostatistician’s analysis and reasoning have made the difference between approval and rejection.
Looking Ahead: New Data Frontiers
Oncology biostatistics has always been complicated, but it faces an ever-growing landscape of complexity moving ahead due to advances in modern clinical research.
- Genomics and biomarkers are subdividing cancers into ever-smaller patient groups, resulting in clinical studies with very small sample sizes. Techniques from rare disease research, such as Bayesian borrowing or external controls, are increasingly applied to oncology.
- Real-world evidence (RWE) offers opportunities for historical comparators, though regulators remain cautious about consistency and bias.
- AI and machine learning promise more efficient data processing and predictive modeling, but regulators demand transparency and explainability.
- Bayesian adaptive methods and model-based drug development are gaining traction, offering ways to make earlier decisions while maintaining rigor.
The Architects of Credibility
Behind every approved oncology drug stands a biostatistician as an integral member of the team that shaped its development pathway. From redefining endpoints to salvaging trial designs to defending evidence before regulators, biostatisticians are the architects of credibility in cancer research. Their contributions may not always be visible to patients, but their impact is profound. Cancer therapies that now extend life and improve outcomes owe much of their success to the biostatisticians who ensured that those therapies were properly tested and proven over the course of their clinical trials.