Conceptual image for prostate cancer treatment: © Dr_Microbe – stock.adobe.com
Metastatic hormone-sensitive prostate cancer (mHSPC), also known as metastatic castration-sensitive prostate cancer, is a type of prostate cancer that has spread beyond the prostate gland to other parts of the body, such as lymph nodes, bones, lungs, or liver. Crucially, it is still responsive to hormonal therapy, meaning treatments that reduce testosterone levels can help slow its growth. Androgen receptor pathway inhibitors (ARPIs) are the guideline-recommended treatment for this disease state.
“When patients have metastatic disease, these ARPIs all improve survival, including enzalutamide (Xtandi),” explained Andrew Armstrong, MD, MSc, a medical oncologist specializing in genitourinary cancers at Duke Cancer Institute. Other approved ARPIs include darolutamide (Nubeqa) and abiraterone (Zytiga).
Direct head-to-head randomized controlled trials (RCTs) comparing the efficacy of enzalutamide and darolutamide do not exist. While RCTs are considered the gold standard for comparing treatments, they can be costly and time-consuming to conduct.
Matching-adjusted indirect comparison (MAIC) is a valuable statistical method used when direct comparisons are unavailable. It allows for indirect comparisons between separate trials by adjusting for differences in baseline patient characteristics and other factors that could influence treatment outcomes, thereby minimizing bias and providing credible efficacy estimates.
MAIC works by using individual patient data from one trial—in this case, the phase 3 ARCHES trial (NCT02677896) for enzalutamide—and matching it to the published aggregate data (PAD) of another trial—here, the phase 3 ARANOTE trial (NCT04736199) for darolutamide. The process involves:
- Identifying study populations: Defining the patient groups from each trial.
- Identifying and selecting effect modifiers: Through literature reviews, statistical analyses, and expert opinion, factors that might alter treatment effects (eg, age, race, disease severity, prior treatment) are identified.
- Reweighting patients: Individual patients in the trial are “upweighted” (given more importance) if they closely match the average patient characteristics in the PAD trial and “downweighted” if they are a poor match. This weighting process, similar to propensity score matching, balances baseline characteristics between the two trial populations.
- Comparing outcomes: After matching, the mean treatment outcomes (like efficacy and safety) are recalculated and compared across the now-balanced trial populations. In this study, a “common comparator” (placebo plus androgen deprivation therapy [ADT]) was used as an anchor to facilitate the indirect comparison.
“These two trials were very similar in some respects,” explained Armstrong. “They both involved men who had metastatic hormone-sensitive prostate cancer who had never been treated before with another AR inhibitor. They were both global studies. They both had the exact same imaging frequency to assess patients’ responses and their PSA outcomes. They had the same end point, which is radiographic progression-free survival [rPFS].”
However, Armstrong noted there were important differences between the studies as well.
“For example, the ARANOTE study did not involve any US patients. It largely occurred in South America, Eastern Europe, and Asia. There were differences in performance status and the proportion of patients who had prior docetaxel chemotherapy. For example, ARCHES had about 17 to 18%, and ARANOTE had 0%. So, these are important adjustment factors. When we did the MAIC, we had to consider all these differences across trials that might confound results,” Armstrong said.
The MAIC study demonstrated that treatment with enzalutamide in combination with ADT was associated with statistically significant improvements in key efficacy outcomes compared to darolutamide plus ADT in patients with mHSPC. Specifically:
- rPFS: Enzalutamide significantly prolonged rPFS, showing a 46% lower risk of radiographic progression or death.
- Time to castration resistance: Enzalutamide significantly extended the time to progression to castration resistance, with a 43% lower risk.
- Other end points: While not statistically significant, outcomes for time to prostate-specific antigen (PSA) progression and time to initiation of new antineoplastic therapy also favored enzalutamide. These results were consistent in both the overall study population and a subgroup of docetaxel-naive patients.
MAIC offers several advantages, especially when direct head-to-head RCTs are not available. They can provide credible efficacy and safety estimates by adjusting for biases introduced by imbalances in trial design or patient baseline characteristics. MAIC specifically corrects for between-trial imbalances in observed effect modifiers. It allows for valuable, tailored insights into treatment efficacy, which can ultimately support better patient outcomes. Additionally, MAIC is a quicker and less expensive way to generate comparative evidence compared to conducting new RCTs, and, unlike some other indirect treatment comparisons, MAIC does not rely on the assumption of similarity and consistency between trials.
“I think this would be provocative data that could set the stage for a future controlled study where you’re testing different ARPIs against each other for efficacy in the real world,” Armstrong said. “Short of that data, which may take 5 or 10 years to generate, we have this matching-adjusted indirect comparison.”
However, like any statistical method, MAIC has limitations. The analysis can only adjust for known and observed effect modifiers. Any unmeasured differences between trial populations could still bias the results. If there’s poor overlap in characteristics between the populations being compared, the effective sample size (ESS) after matching can be significantly reduced. In this study, the ESS for ARCHES was reduced by 72% after matching due to differences in region, race, and ECOG PS between the ARCHES and ARANOTE populations.
Additionally, MAIC may not be able to adjust for certain factors. For example, this study could not adjust for significant differences in median baseline PSA due to confounding by ADT timing, nor could it compare safety outcomes due to a lack of exposure-adjusted safety data from one of the trials.Further, biases may remain due to inherent differences between trials (eg, tumor genetics, study timeframe) that cannot be fully adjusted for. And while assessed as plausible in this study, the validity of this assumption is crucial for time-to-event analyses.
Given that multiple ARPI options are recommended by clinical guidelines for mHSPC, the findings from this MAIC study provide important new knowledge for health care professionals. By demonstrating the statistically significant superior efficacy of enzalutamide over darolutamide in slowing disease progression, reducing the risk of radiographic progression or death, and delaying progression to castration resistance, these results can help inform shared decision-making around treatment for patients with mHSPC, alongside other patient-specific factors.
Armstrong noted, however, that efficacy is not the only factor that comes into treatment decisions.
“The cost of these drugs, the availability of the drugs, and the toxicities and safety for your patient in front of you. That individual patient may have different comorbidities and preferences and toxicities based on their age and drugs that they’re taking,” Armstrong said. “[Efficacy] is one part of the equation.”