Classifying multiple lung cancers using morphological features: a meta-analysis

Multiple lung cancers may be intrapulmonary metastases or multiple primaries. Management and prognosis of both entities are different and make the pathologist's role challenging. In fact, distinguishing both entities may be difficult especially when the tumors present the same microscopic subtype. The microscopic diagnoses of these tumors have been improved based on the 2015 WHO classification. The aim of the authors was to assess the diagnostic value of morphologic features in comparison to the gold standard test represented by molecular testing in the distinction between intrapulmonary metastases and multiple lung primaries. To retrieve all eligible articles, PubMed and Embase databases and Cochrane Library were comprehensively searched from 1999 to 2020 with limitation to French andEnglish language. The Meta-Disc software 5.1.32 was used to conduct this meta-analysis. The pSEN, pSPE, NLR, PLR, and DOR with the 95% confidence intervals were calculated. The area under the SROC was calculated based on the SEN and SPE of each study. Q test and I² statistics were carried out to explore the heterogeneity among studies. P value <.1 for q test or I² value >50% represented substantial between-study heterogeneity. Meta-regressions were performed to explore the sources of heterogeneity if necessary. Twelve eligible articles with 309 patients were included. pSEN was estimated to 65% with I-square estimated to 53%. pSPE reached 49% with I-square estimated to 56%. PLR was estimated to 1.23 with I-square estimated to 33%. NLR was estimated to 0.65 with I-square estimated to 23.1%. dOR was estimated to 2.13 [1.07-4.25] with I-square estimated to 26.5%. AUC was estimated to 0.63. The meta-regression analysis showed non-significant effect of the WHO classification, next-generation sequencing, or nucleotide-specific sequencing with p reaching respectively 0.38, 0.06, and 0.36. These results highlight that morphologic features may be useful in the diagnosis of multiple lung cancers especially when dealing with surgical specimen. The mild heterogeneity observed in this meta-analysis may be due to other covariates that were not described in the different articles including the sample nature.