AUA 2018: Factors Associated with Early Mortality and Survival Outcomes Among Patients with Testicular Neoplasms

San Francisco, CA ( The group from Princess Margaret Cancer Centre in Toronto, Canada presented their >4-decade population-based study assessing predictors of early mortality among patients diagnosed with testicular neoplasms. Since Dr. Lawrence H. Einhorn’s landmark chemotherapy studies of the 1970’s and 1980’s that remarkably improved survival among testis cancer patients, the long-term survival rates generally exceed 95%.  However, despite these extraordinary survival outcomes, there is a small subset of patients that will still suffer mortality from testis cancer. Zach Klaassen, MD, and colleagues sought to assess factors associated with “early” mortality of any cause after testicular cancer diagnosis using a population-based cancer registry. Secondary objectives were to assess predictors of cancer-specific mortality (CSM) and overall survival (OS).

The Surveillance, Epidemiology, and End Results (SEER) database was queried from 1973-2011 to identify men with a testicular neoplasm. Because the primary outcome was early mortality within 2-years of diagnosis, all patients had at least 2 years of follow-up to be eligible for the study. Since >95% of patients with testis cancer have germ cell tumors (seminoma or non-seminomatous germ cell tumor (NSGCT)), the authors performed a sensitivity analysis limited to these patients. Covariates included: age, race (white, black, other), marital status (married vs unmarried), USA region (south, west, midwest, northeast), laterality, RPLND (yes/no), histology (seminoma, NSGCT, lymphoma, variant), radiation (yes/no), chemotherapy (yes/no), and SEER stage (localized, regional, distant). A multivariable log binomial regression model was used to generate prevalence ratios (PR) for predictors of early mortality. For the survival analysis, a multivariable competing risks analysis was used to generated hazard ratios (HRs) for identifying factors associated with CSM (censoring for other cause mortality), and a multivariable Cox-proportional hazards model was used to generate HRs for identifying factors associated with OS.

The majority of patients had seminoma (54.1%) or NSGCT (41.7%), whereas the rest included lymphoma (3.0%) and variant histology (1.2% - mostly sarcomas). The mean age was 35.4 (SD 12.4) years, 92% of patients were white, and 49.1% of patients were married at the time of diagnosis. 35% of patients received radiation, and 30% received chemotherapy over a median follow-up of 134 months (IQR 74-228). 5.1% of patients died ≤2 years after diagnosis, 6.1% of patients had cancer specific mortality, and 8.1% suffered all-cause mortality. 

Predictors of early mortality included (model AUC 0.875):

  • Histology: vs seminoma – NSGCT (PR 1.59), lymphoma (PR 4.01), variant histology (PR 6.36), all p<0.001
  • Race: black vs white (PR 1.96, p=0.02)
  • Disease burden: vs localized – regional (PR 2.73), distant (PR 4.51), both p<0.001
  • Married: PR 0.72 (p<0.001)
  • Later years of diagnosis: vs 1973-1992 – 2003-2011 (PR 0.31, p<0.001)
  • RPLND: PR 0.27 (p=0.001)
Predictors of CSM included age (HR 1.056, 95%CI 1.054-1.059), black race (vs white HR 1.63, 95%CI 1.42-1.88), NSGCT (vs seminoma HR 1.69, 95%CI 1.56-1.83), lymphoma (vs seminoma HR 2.33, 95%CI 1.90-2.87), variant histology vs seminoma (HR 2.81, 95%CI 2.39-3.30), married (HR 0.61, 95%CI 0.58-0.65), RPLND (HR 0.45, 95%CI 0.35-0.57), SEER stage regional (vs localized HR 1.54, 95%CI 1.43-1.66), and SEER stage distant (vs localized HR 4.46, 95%CI 4.09-4.88). Similar predictors were reported for assessing OS and the subgroup analysis restricted to patients with NSGCT or seminoma histology. Radiation therapy and chemotherapy as time-varying covariates were not significant in the final OS model (HR 1.48, 95%CI 0.89-2.47; HR 0.92, 95%CI 0.55-1.52).
Given the low mortality event rate, the strength of this study is the nearly four-decade population-level analysis to enable multivariable models assessing early predictors for mortality, and long-term survival outcomes. However, limitations noted by the authors include the lack of comorbidity data and information regarding chemotherapy side-effects and/or complications. The authors concluded that although patients with testicular cancer have excellent long-term survival, there is a subset of patients that suffer mortality within 2 years of diagnosis. Similar to what has been reported for other malignancies, unmarried and black patients are more likely to suffer early mortality, raising the question of appropriate access to initial and follow-up care.

Presented by: Zachary Klaassen, MD University Health Network, Princess Margaret Cancer Centre, Toronto, ON, Canada
Co-Authors: Thenappan Chandrasekar, Hanan Goldberg, Toronto, Canada, Karan Arora, St. George's, Grenada, Rashid K. Sayyid, J. Charles Victor, Neil E. Fleshner, Robert J. Hamilton, Girish S. Kulkarni, Toronto, Canada

Written by:  Zachary Klaassen, MD, Urologic Oncology Fellow, University of Toronto, Princess Margaret Cancer Centre at the 2018 AUA Annual Meeting - May 18 - 21, 2018 – San Francisco, CA USA