ORLANDO – Evidence from randomized clinical trials is all over the map, but a new study using statistical models to compare patient outcomes in a population-based cohort indicates that adjuvant chemotherapy can significantly improve overall survival in patients with locally advanced bladder cancer.
Using various propensity-score models to minimize the impact of confounding when comparing results across a large population sample, Dr. Matthew Galsky and his colleagues from the Icahn School of Medicine at Mount Sinai, New York, found that adjuvant chemotherapy was associated with improvements in overall survival from 28% to 38%, compared with an observation cohort of patients who did not receive adjuvant chemotherapy.
Dr. Matthew Galsky
The results, based on data from 5,653 patients with stage pT3 or greater transitional cell carcinoma of the bladder with or without positive nodes but no distant metastases, are similar to those seen in smaller meta-analyses, Dr. Galsky said at the 2015 Genitourinary Cancers Symposium sponsored by the American Society of Clinical Oncology.
“In this large population-based analysis, adjuvant chemotherapy was associated with improved overall survival. The effect size was similar to that which has been reported in the meta-analyses of randomized trials, and importantly, these are data from a real-world population, perhaps providing some insights into the effectiveness of this approach,” he said.
Previously published randomized trials comparing chemotherapy regimens ranging from single-agent cisplatin to combination regimens such as MVAC (methotrexate, vincristine, doxorubicin and cisplatin) have differed in population size and outcome, with three of eight trials finding a survival benefit for adjuvant chemotherapy, but the remainder showing no benefit.
“Early trials of adjuvant chemotherapy were critical in establishing the feasibility of this approach, yet in retrospect utilized suboptimal chemotherapy regimens, were underpowered, and at times were associated with methodologic flaws. Three recent trials have compared contemporary chemotherapy regimens administered in the adjuvant setting with observation in patients with locally advanced bladder cancer post cystectomy. Unfortunately, all three of these trials closed early due to poor accrual, enrolling only 39% of the planned subjects,” Dr. Galsky said.
To overcome the problem of a lack of large randomized trials, the investigators created a population-based study of patients with pT3 and/or pN+, M0 bladder cancer who underwent radical cystectomy from 2003 through 2007. The patients, identified from the National Cancer Data Base, did not receive either neoadjuvant chemotherapy or radiation to the primary tumor. Patients who did not receive adjuvant chemotherapy and survived more than 30 days after cystectomy were chosen for the observation arm.
Minimizing bias
Propensity score matching generates the conditional probability of one individual being treated with a particular treatment approach given multiple pretreatment covariates. This method allows investigators to balance covariates between the two groups, with the goal of eliminating or minimizing treatment allocation biases that may can affect the relationship between the treatment and postoperative outcomes.
The authors used logistic regression to calculate propensity scores that represented the predicted probabilities of assignment to adjuvant chemotherapy rather than observation based on factors that included age, demographics, year of diagnosis, tumor stage, surgical margin status, lymph node density, distance to the hospital, number (volume) of cystectomies performed at the hospital, and hospital type and location.
They matched every patient assigned to adjuvant therapy to two controls from the observation group. They also created propensity score and inverse probability of treatment-weighted proportional hazard models to estimate adjusted hazard ratios for overall survival among all patients in the sample.
Finally, they conducted a sensitivity analysis to look at the impact of poor performance status on survival, and subset analyses of the effects of age, gender, and lymph node status.
A total of 5,653 patients were included in the sample, 4,360 of whom underwent observation, and 1,293 of whom received adjuvant chemotherapy.
The effect of adjuvant chemotherapy on overall survival was fairly consistent across propensity score–adjusted models. The hazard ratio for adjuvant chemotherapy in a model with stratification by performance status quintile was 0.72. In a propensity score–weighting model, looking at the inverse probability of treatment weighting, the HR was also 0.72. Finally, in a propensity score–matching model, the HR for adjuvant chemotherapy vs. observation was 0.62 (P < .0001 for all).
They found that patients who received adjuvant chemotherapy were significantly more likely to be younger, have more lymph node involvement, have higher tumor stage, positive margins, reside in the Northeast and closer to the hospital, and to have private insurance.
The findings were supported by sensitivity analyses controlling for performance status and comorbidities.
Dr. Galsky noted that the study was limited by its retrospective design, lack of details about chemotherapy, and a lack of data on recurrence or cancer-specific survival.