• 2019-07
  • 2019-08
  • 2019-09
  • 2019-10
  • 2020-03
  • 2020-07
  • 2020-08
  • 2021-03
  • Kaplan Meier analyses for PFS and OS were conducted


    Kaplan-Meier-analyses for PFS and OS were conducted according to the relative change in the leading STM (decrease, increase<2-fold, increase≥2-fold) as well as to radiological change using RECIST criteria (CR/PR, SD, PD). Results were expressed as median in months (M) and 95% confidence interval unless otherwise specified. The resulting survival curves were compared statistically using the log rank test, whereas a p-value<0.05 was regarded statistically significant. Evaluation of predictive factors for progression/death and death was conducted applying univariate and step-wise multivariate Cox-regression analyses. Variables analyzed in these models were age in groups (<70, ≥70 years), sex, smoking status (<, ≥5 pack years), histological subtype (adeno-, squamous-cell carcinoma), presence of APTSTAT3-9R metastases, palliative therapy line (1/2, ≥3), TNM stage (III, IV) [17], ECOG (0, 1/2) and presence of a targetable mutation. Co-morbidities were assessed by the Charlson Comorbidity Index (CCI) [18], whereas a cutoff value ≥3 was determined for separation of patients with little versus extensive burden of comorbidity [[19], [20], [21]]. Additionally, PD-L1 status (positive, negative) assessed with a 22C3 assay for Autostainer Link 48 by Dako (Agilent Technologies, Santa Clara, CA), STM response (decrease, increase <2-fold and ≥2-fold) and RECIST response (CR/PR, SD/PD) were included.
    Results Eighty-four patients with ICI treatment initiation between August 2015 and May 2018 met the predefined requirements to be included in the analysis. Baseline patient and tumor characteristics are shown in Table 1. Mean STM baseline values and STM dynamics are depicted in Table 1, Table 2. Most frequently identified leading STM were CEA (53%) and CYFRA 21-1 (38%). CA19-9 was leading STM in 8% of cases, NSE in one patient (1%). This individual also displayed an elevated CYFRA 21-1 level at baseline decreasing in parallel with NSE upon the first restaging. In eight cases, no STM analyzed was elevated at the time of ICI initiation. Among those, in three patients with a leading STM of CEA and no other available STM, CEA increased within the normal range in 2 cases (best radiological response for both: PD) and decreased in one case (best radiological response: SD). Of 5 patients with non-elevated CYFRA 21-1 as leading STM, there were four consecutive increases above the upper limit of normal (best response: 1 SD, 1 PR, 2 PD) and one decrease (best response: PD). Two of cytokinins cases had concomitant CEA dynamics also within the normal range. ROC analyses revealed the best predictive power of baseline STM levels for both death and progression/death for CYFRA 21-1 (AUC death: 68.9%, progression/death: 71.8%) and for CA19-9 (AUC: 64.7% and 67.6%), followed by CEA (AUC: 54.6% and 51.1%), and NSE (AUC: 53.2% and 44.1%). In terms of STM dynamics, CEA (AUC death: 67.7%, progression/death: 64.7%) and CYFRA 21-1 (AUC: 60.4% and 58.1%) were superior predictors as compared to CA19-9 (AUC: 53.9% and 48.1%). Concerning NSE, the number of measurements was too small for a valid ROC analysis of STM dynamics. The model of a leading STM surpassed the power of single STM analyses for STM dynamics (AUC death 70.3%, progression/death: 72.3%) but not for baseline STM (AUC: 60.4% and 57.2%). (Supplementary Tables 1–3, supplementary Figures 1–4) Both median PFS and OS significantly (p < 0.001) differed between the specified STM and RECIST response categories (Table 2, Fig. 1). When separating individuals with radiological responses (PR/CR) from stable or progressive disease (SD/PD), both groups showed a significantly (p < 0.001) better median PFS and OS, when STM concurrently decreased. In several well-responding subgroups (CR/PR + STM decrease, CR/PR + STM decrease and increase), median PFS and OS could not be calculated, as the number of patients having progressed or died was too small. (Table 3, Fig. 2)