OBJECTIVES:Extrapolating survival data beyond the duration of a clinical trial is increasingly being used to provide long-term survival benefits in cost-effectiveness analysis of new cancer drugs. However, access to individual patient data (IPD) is rarely available, and thus researchers need to reconstruct IPD from published Kaplan-Meier survival curves, so that parametric survival analysis can be conducted. This study aims to reconstruct IPD from a case match control study by O’Connor et al. (2018), which compared the overall survival outcome of pralatrexate to historical controls in patients with peripheral T-cell lymphomas.

METHODS:The Kaplan-Meier survival curves for both groups were digitized using DigitizeIt software. The method developed by Guyot et al. (2012) was applied to reconstruct IPD from the extracted coordinates, using R software. Guyot’s algorithm is the most commonly used method and closely approximates the original survival curves if numbers at risk and total number of events are reported. The estimated median overall survival and hazard ratio (HR) of the reconstructed data were compared to the original ones to assess the validity of results.

RESULTS:Based on the reconstructed IPD, 50 and 68 events occurred in the pralatrexate and control groups, respectively. The estimated median overall survival of pralatrexate and control groups was 15.40 and 4.10 months, respectively, which was similar to the results in the O’Connor publication (15.24 and 4.07 months). Similarly, compared with HR reported by the original publication (HR: 0.432, 95% confidence interval (CI): 0.298-0.626), the reconstructed IPD showed a HR of 0.431 (95% CI: 0.298-0.623), demonstrating the validity of the reconstructed data.

CONCLUSIONS:Generating patient-level data from the published survival curves is feasible by adopting established methods. The reconstructed IPD can be utilized for fitting parametric survival models. This, in turn, can provide transition probabilities which are necessities for cost-effectiveness analysis of pralatrexate in patients with peripheral T-cell lymphomas.