Neuroanatomical location of lung cancer brain metastases in 234 patients with a focus on cancer subtyping and biomarkers

Abstract

Brain metastases are frequent in neuropathology practices; however, the literature on their distribution is frequently derived from imaging studies. This work examined metastases of lung cancer to the brain through the lens of pathology specimens. All brain surgical pathology cases accessioned from 2011-2020 were retrieved from a regional laboratory. Specimens were classified by neuroanatomical location, diagnostic category, and diagnosis with a hierarchical free text string-matching algorithm. All reports classified as probable metastasis per algorithm were reviewed by a pathologist. Lung biomarkers and selected immunostains were retrieved with text parsing and reviewed. Among 4,625 cases of brain surgical resection specimens, 854 were classified as probable metastasis by the algorithm. On report review, 538/854 cases were confirmed as metastasis with a known primary site. The 538 cases were from 511 patients and 234/511 patients had lung primaries. Small cell lung cancer lesions were most frequently found in the cerebellum (17/30). Lesions from lung adenocarcinoma (59/164) and non-small cell carcinoma-not otherwise specified (NSCLC-NOS) (15/34) were most commonly found in the frontal lobe. Squamous cell carcinoma lesions were most commonly found in the frontal and occipital lobes (8/27). 72/234 cases were reported as NSCLC-NOS and could be further subclassified using immunostaining (41/72). Lung biomarker data were retrieved in ~38% of cases. PD-L1 positivity was dependent on neuroanatomical distribution (p = 0.04); other examined biomarkers were not. The distribution of lung tumours metastatic to the brain is dependent on the lung cancer subtype (p<0.001). The reporting of histologic subtype could be further optimized in the local environment.

Fig 1. The study methods in a flow chart.

HFTSMA = hierarchical free text string-matching algorithm; csv = formatted text output (tab or comma separated).

Fig 2. The brain metastases from the lung by neuroanatomical site.

This figure is applicable in the diagnostic context; if the neuroanatomical site is known, it answers the question “How likely is a metastasis from a lung tumour of subtype ‘X’?”.

Fig 3. The brain metastases by lung subtype.

This figure is applicable in the pathobiologic context; if the lung subtype is known, it answers the question “How likely is a metastasis to the neuroanatomical site ‘Y’?”.

Fig 4. Clustered heatmap showing the p-values calculated by Fisher’s exact test for all each pairing of neuroanatomical location and lung cancer subtype.

Neuroanatomical location abbreviations: FRO = frontal lobe, PAR = parietal lobe, TEM = temporal lobe, OCC = occipital lobe, CER = cerebellum. Lung cancer subtype abbreviations: AC = adenocarcinoma, SCC = squamous cell carcinoma, SCLC = small cell carcinoma, NSCLC = non-small cell carcinoma-NOS. The colours represent different p-value ranges: red is p = 0 to <0.05, pink is p = 0.05 to <0.1, white is p = 0.1 to <0.2, light blue is p = 0.2 to <0.4 and dark blue is p = 0.4 to 1.0.