Potential utility of albumin–bilirubin and body mass index-based logistic model to predict survival outcome in non-small cell lung cancer with liver metastasis treated with immune checkpoint inhibitors

Le résumé fourni par la source

To the Editor: Lung cancer is characterized by its high incidence and mortality rates, imposing a significant burden of disease globally.[1] Liver metastasis is associated with poorer clinical outcomes and shorter life expectancy in non-small cell lung cancer (NSCLC).[2] Although immune checkpoint inhibitors (ICIs) have revolutionized the treatment of NSCLC, existing predictive and prognostic biomarkers have limited effectiveness and pose a clinical challenge, particularly for patients with liver metastasis. The albumin–bilirubin (ALBI) grading system, an evidence-based measure of liver dysfunction calculated from blood albumin and bilirubin levels, has shown promise in predicting prognosis and ICI efficacy in NSCLC.[3] Nonetheless, high body mass index (BMI) was independently associated with improved overall survival (OS) with ICI therapy in NSCLC.[4] Given that the prognosis is worse for patients with liver metastasis and low BMI despite receiving ICI treatment, we proposed that assessing both the ALBI score and BMI together could offer valuable prognostic information for patients with NSCLC and liver metastasis who are undergoing ICI therapy. This retrospective multicenter study assessed the utility of ALBI-B, a logistic regression-based model that integrates pretreatment ALBI score and BMI, in predicting clinical outcomes with ICIs for patients with NSCLC and liver metastases. This study has been approved by the Institutional Review Board of Hunan Cancer Hospital (No. 2019-SSB-IIT-120) and encompassed patients who were administered an ICI-containing regimen (regardless of treatment line) following the identification of liver metastases derived from NSCLC. The requirement to obtain the informed consent was waived. The ALBI score was calculated based on pretreatment serum albumin and bilirubin levels using the formula: ALBI score = (0.66 × log10[bilirubin μmol/L]) – (0.085 × [albumin g/L]).[3,5] BMI was calculated as BMI = (weight kg)/(height m)2. Supplementary Figure 1A, https://links.lww.com/CM9/C221 summarizes the patient screening flow chart. Cohort A was retrospectively screened from 3475 patients diagnosed with NSCLC between January 1, 2018 and February 8, 2022. Supplementary Table 1, https://links.lww.com/CM9/C222 lists the baseline characteristics of the 110 patients included in Cohort A. Cohort A had a median progression-free survival (PFS) of 5.0 months (95% confidence interval [CI]: 4.1–5.9) and a median OS of 17.0 months (95% CI: 10.2–23.8) [Supplementary Figure 2, https://links.lww.com/CM9/C221]. PFS and OS were calculated from the start of ICI treatment until the respective endpoint. High BMI was associated with favorable PFS with ICI in patients with NSCLC and liver metastases in both univariable (hazard ratio [HR] = 0.89, 95% CI: 0.84–0.95; P <0.001) and multivariable analyses (HR = 0.89, 95% CI: 0.83–0.95; P <0.001, Supplementary Figure 3, https://links.lww.com/CM9/C221). Contrastingly, high ALBI score was identified as a predictor of unfavorable PFS with ICI in both univariable (HR = 2.91, 95% CI: 1.87–4.51; P <0.001) and multivariable analyses (HR = 3.95, 95% CI: 2.35–6.64; P <0.001, Supplementary Figure 3, https://links.lww.com/CM9/C221). Moreover, ALBI score and BMI were still associated with PFS even when the multivariable analysis only included the statistically significant variables from the univariable analysis [Supplementary Table 2, https://links.lww.com/CM9/C222]. Variance inflation factor assessment also showed low multicollinearity of the variables [Supplementary Table 3, https://links.lww.com/CM9/C222]. Patients with programmed death-ligand 1 tumor proportion score (PD-L1 TPS) ≥50% had a PFS advantage over patients with PD-L1 TPS <1% based on multivariable analysis (P = 0.013, Supplementary Figure 3, https://links.lww.com/CM9/C221). Patients treated with the ABCP (atezolizumab + bevacizumab + carboplatin + paclitaxel) or chemotherapy plus ICI demonstrated substantially prolonged PFS compared with ICI monotherapy in multivariable analysis (Chemo + ICIs vs. PD-1 monotherapy, P = 0.029; ABCP vs. PD-1 monotherapy, P = 0.001, Supplementary Figure 3, https://links.lww.com/CM9/C221). Regarding OS, only high ALBI score was identified as a factor for unfavorable prognoses in both univariable (HR = 2.28, 95% CI: 1.37–3.78; P = 0.001) and multivariable analyses (HR = 3.03, 95% CI: 1.63–5.64; P <0.001) [Supplementary Figure 4, https://links.lww.com/CM9/C221, Supplementary Tables 4 and 5, https://links.lww.com/CM9/C222]. ALBI-B, a prediction model combining ALBI score and BMI, was established based on a logistic regression model using the formula ALBI-B = −7.2043 – ([0.9406 × ALBI score] + [0.2084 × BMI]). The median PFS of Cohort A served as the reference value for constructing the logistic regression models. The receiver operator characteristic (ROC) curve shown in Supplementary Figure 1B, https://links.lww.com/CM9/C221 shows that the predictive performance of ALBI-B was better than the individual assessment of either ALBI score or BMI. The ALBI-B model yielded an area under the curve (AUC) of 0.776 (95% CI: 0.686–0.850), with 83.3% sensitivity and 67.7% specificity for predicting PFS outcomes with ICI in Cohort A, which outperformed PD-L1 TPS (AUC = 0.562, 95% CI: 0.398–0.726; Supplementary Figure 1B, https://links.lww.com/CM9/C221). Cohort A was stratified as ALBI-B-Low and ALBI-B-High, using an optimal cutoff of −0.477 as determined by the Youden index of 0.511. Compared with the ALBI-B-Low subgroup, the ALBI-B-High subgroup had significantly longer PFS (7.0 vs. 3.0 months; P <0.0001; HR = 2.82, 95% CI: 1.78–4.48) and OS (27.0 vs. 9.0 months; P <0.0001; HR = 2.71, 95% CI: 1.57–4.69) [Supplementary Figure 1C, https://links.lww.com/CM9/C221]. We also investigated the relationship between tumor response and the ALBI-B model. The best objective response data in Cohort A were as follows: 0 complete response (CR), 32.7% (n = 36) partial response (PR), 40.0% (n = 44) stable disease (SD), and 27.3% (n = 30) progressive disease (PD) [Supplementary Figures 5 and 6, https://links.lww.com/CM9/C221]. ALBI-B-High subgroup had a significantly higher objective response rate (ORR, 48.3% vs. 14.0%, P <0.001) and disease control rate (DCR, 88.3% vs. 54.0%, P <0.0001) than the ALBI-B-Low subgroup [Supplementary Figure 7, https://links.lww.com/CM9/C221]. Subgroup analyses revealed that regardless of the treatment regimens received (ICI monotherapy vs. atezolizumab/bevacizumab/carboplatin/paclitaxel [ABCP] vs. chemotherapy plus ICI), treatment setting (first-line vs. ≥second-line), or presence/absence of actionable genetic mutation, the ALBI-B-High subgroup had significantly longer PFS than the ALBI-B-low subgroup [Supplementary Figure 8, https://links.lww.com/CM9/C221]. A subgroup analysis of patients with oncogenic driver gene-positive NSCLC who received ICI as subsequent-line therapy revealed better disease control and longer PFS for the ALBI-B-High subgroup than in the ALBI-B-Low subgroup [Supplementary Figure 9, https://links.lww.com/CM9/C221]. To validate the predictive ability of the ALBI-B model, we used the clinical data of an independent cohort of 60 patients with NSCLC liver metastasis (Cohort B) derived from screening 1000 patients diagnosed with NSCLC between 2018 and 2022 who were enrolled in prospective clinical trials for ICIs, including Orient 11/12/31, Camel/sq, Rational 303/304/307, Choice 01, Impower 150, and Checkmate 078. Cohort B had similar baseline clinicopathological characteristics as Cohort A, with ALBI-B-High and ALBI-B-Low subgroups having significantly different median serum albumin, median ALBI score, and median BMI [Supplementary Table 6, https://links.lww.com/CM9/C222]. Cohort B had a median PFS of 6.0 months (95% CI: 5.3–6.7) and a median OS of 13.7 months (95% CI: 10.5–16.9) [Supplementary Figure 10, https://links.lww.com/CM9/C221]. Using the same ALBI-B cut-off value of −0.477, Cohort B was stratified into AL

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.