The authors reported a pooled prognostic evaluation of the CALLY index—an integrated marker combining C-reactive protein, albumin, and lymphocyte count—across primary lung malignancies. They searched Embase, PubMed, Scopus, and Web of Science up to 18 September 2026. Eleven studies provided data for the overall survival analysis, contributing 13 cohorts; seven studies provided data for combined recurrence/progression-free survival (RFS/PFS). The preprint states that all included data were from published sources and that the review itself has not undergone peer review.
Data indicated that a lower CALLY index was associated with inferior overall survival. The pooled hazard ratio for lower versus higher CALLY was 1.78 (95% CI 1.49–2.12). The study team observed substantial between-study heterogeneity for OS, with I2 reported at 71.6%. The investigators used random-effects models with restricted maximum likelihood estimation and applied the Hartung–Knapp adjustment to account for uncertainty in pooled estimates. Funnel-plot asymmetry was present for OS, and the authors interpreted this as possible small-study effects.
For combined recurrence/progression-free survival outcomes, the pooled association followed the same direction: lower CALLY predicted worse RFS/PFS with an HR of 1.60 (95% CI 1.40–1.84). Heterogeneity for these endpoints was reported as I2 = 0%, indicating consistency across the included cohorts for RFS/PFS. The authors noted that leave-one-out sensitivity checks did not materially change the pooled RFS/PFS estimate.
Investigators appraised risk of bias using the Quality In Prognosis Studies (QUIPS) tool. Seven studies were rated as having moderate overall risk of bias, while four were rated high. Using GRADE, the authors downgraded the certainty of evidence to low for both OS and RFS/PFS, citing risk of bias, heterogeneity (for OS), and endpoint indirectness as principal concerns. The presence of funnel-plot asymmetry prompted caution about small-study effects potentially inflating pooled estimates.
The review team conducted sensitivity analyses, including leave-one-out procedures, and reported that the principal associations remained. Subgroup analyses and meta-regression were performed to probe sources of heterogeneity; the authors reported no statistically significant subgroup differences. The preprint does not provide granular subgroup effect estimates or detailed meta-regression coefficients within the abstract, and the main text should be consulted for those specifics.
The authors identified several methodological limitations that restrict interpretation. Measurement approaches and cutoff definitions for the CALLY index varied across studies, and endpoint definitions (for example, how recurrence or progression was ascertained) were not uniform. The review team emphasized that these variations contribute to indirectness and risk of bias. They called for prospective validation using standardized biomarker measurements and clinically coherent outcome definitions to determine whether the CALLY index adds incremental prognostic value beyond established clinical factors. The manuscript also cautions that the pooled estimates derive from studies with moderate-to-high risk of bias and that findings should not yet be used to guide clinical practice because this is a preprint that has not been peer reviewed.
Specific unanswered questions the authors flagged include: what standardized assay or cut-point should be adopted for the CALLY index in lung cancer cohorts, whether CALLY provides independent prognostic information in multivariable models with staging and performance status, and whether the index predicts treatment-specific outcomes (surgery, systemic therapy, or radiotherapy). The authors recommended prospective cohorts or harmonized pooled analyses to resolve these issues.
The study team declared no competing interests and stated that all data used were publicly available from the cited studies. The synthesis ends by calling for prospective, standardized validation rather than asserting clinical readiness; that recommendation is the final actionable message in the preprint.