Trend and socioeconomic disparities in survival outcome of metastatic melanoma after approval of immune checkpoint inhibitors: a population-based study

Abstract

Background: The approval of immune checkpoint inhibitors (ICI) for metastatic melanoma in 2011 has changed the treatment landscape of this disease. However, current trend of the population-based survival remains unclear.

Methods: 8078 patients with metastatic melanoma diagnosed in the pre-ICI (2005-2010) and post-ICI period (2011-2016) were enrolled from the Surveillance, Epidemiology, and End Results (SEER) program for survival comparison. Propensity score matching (PSM) was performed to reduce selection bias. Cox proportional hazards model was applied for identifying survival-related factors and constructing a prognostic nomogram. The accuracy of the nomogram was determined by concordance index (C-index), calibration curves, and validated by an internal cohort.

Results: Patients in the post-ICI period had a significantly longer median overall survival (OS) than those in the pre-ICI period, even after performing PSM between the two periods. We also found socioeconomic disparities in the survival improvement. Significant differences in OS between the two periods were only observed in cases with medical insurance and patients living in urban or low-poverty area, but not uninsured cases and patients from rural or high-poverty area. For patients in the post-ICI period, multivariate analysis demonstrated that socioeconomic and insurance status were independent prognostic factors, which can be combined with other clinical variates into a nomogram for OS prediction with promising C-index of 0.672 and 0.650 in the training- and testing cohort, respectively.

Conclusion: An overall trend to favorable survival at the population level and socioeconomic disparities in the survival trend are observed in metastatic melanoma after the ICI approval. The proposed nomogram is available for prognostication in the current melanoma management.

Keywords: Metastatic melanoma; SEER; immune checkpoint inhibitors; nomogram; survival.

Figure 1

Flow diagram of patient selection.

Figure 2

Survival trend in patients with metastatic melanoma in the SEER database. Kaplan-Meier curve comparing overall (A) and disease-specific survival (B) between the 2005-2010 and 2011-2016 periods. (C) Kaplan-Meier curve comparing overall survival among the four 3-year periods (2005-2007, 2008-2010, 2011-2013, and 2014-2016). (D) 1-, 2-, and 3-year relative survival rates from 2005 to 2016.

Figure 3

Survival comparison after performing propensity score matching between the period year 2005-2010 and 2011-2016. Kaplan-Meier curve comparing overall survival between the 2005-2010 and 2011-2016 periods (left) and among the four 3-year periods 2005-2007, 2008-2010, 2011-2013, and 2014-2016 (right).

Figure 4

Stratification analyses for the trends of survival in metastatic melanoma. Kaplan-Meier plots comparing overall survival for patients ≤65 or >65 years of age (A, B), non-Hispanic white or other races/ethnics including Hispanic, black, Asian and others (C, D), cutaneous or mucosal melanoma (E and F), and primary or secondary melanoma (G, H).

Figure 5

Socioeconomic disparities in the survival improvement of metastatic melanoma. Kaplan-Meier plots comparing overall survival for patients living in urban or rural area (A, B), patients living in low- or high-poverty area (C, D), and patients with medical insurance or cases with no insurance and unknown insurance record (E, F).

Figure 6

Nomogram and calibration curves for predicting overall survival in metastatic melanoma based on the SEER data, 2011-2016. A. In the nomogram, a Points line is drawn upward to measure the prognostic score of each variable, and the sum of these scores is located on the Total Points axis and then projected on the bottom scale to determine the likelihood of 2-, 3- and 5-year overall survival (OS). B, C. Calibration curves in the training set for prediction of 2- and 3-year OS. D, E. Calibration curves in the testing set for prediction of 2- and 3-year OS.