The landscape of small cell lung cancer metastases: Organ specificity and timing

Abstract

Background: Early metastasis is a hallmark of small cell lung cancer (SCLC). However, the mechanisms and resulting patterns of SCLC dissemination are unclear. Our aim was thus to investigate the organ specificity and timing of blood-borne metastases in a comprehensive large cohort of SCLC patients.

Methods: In this retrospective non-interventional cross-sectional study of 1009 Caucasian SCLC patients, we investigated the correlation between the distinct locations of the primary tumor and metastatic sites.

Results: The onset of bone (p < 0.001), brain (p < 0.001), and pericardial (p = 0.02) metastases were late events, whereas adrenal gland (p = 0.005) and liver (p < 0.001) metastases occurred earlier. No significant difference was found in the distribution of early versus late metastases when comparing central and peripheral primary tumors. Patients with bone metastases had a higher than expected likelihood of having liver metastases, while brain metastases tended to appear together with adrenal gland metastases. Pleural and both lung and pericardial metastases also tended to co-metastasize together more frequently than expected if metastatic events occurred independently. Notably, patients with central primary tumors had decreased median overall survival (OS) compared to those with peripheral tumors, although this tendency does not appear to be significant (p = 0.072).

Conclusion: Our results are suggestive for particular site- and sequence-specific metastasis patterns in human SCLC. SCLC bone metastases tend to appear together with liver metastases, while brain metastases occur together with adrenal gland metastases. Better understanding of metastasis distribution patterns might help to improve the diagnosis and therapeutic decision-making in SCLC patients.

Keywords: metastasis pattern; small cell lung cancer; survival.

FIGURE 1

Metastatic site‐specific time sequence in 1009 SCLC patients. (a) The onset of bone (**p < 0.001), brain (**p < 0.001), and pericardial (**p = 0.02) metastases tended to be late, whereas the development of adrenal gland (p* = 0.005) and liver (*p < 0.001) metastases were usually early events during SCLC progression (*favors early metastases, **favors late metastases, Chi‐square test with Bonferroni‐correction). (b) There was no statistically significant difference in the distribution of early vs. late metastasis appearance when comparing centrally and peripherally located primary SCLCs. (c) The percentage of SCLC patients with metastases in each organ according to the corresponding primary tumor's bronchoscopic localization (central or peripheral)

FIGURE 2

Metastatic patterns and interarrival times before and after the appearance of SCLC metastases in specific organs. (a) The number in the boxes represents the total number of patients with metastases in both organs i (x axis) and j (y axis). Based on a gradient scale from blue to yellow, combinations with weak association are highlighted in blue, whereas those with strong association are highlighted in yellow. The number of co‐occurrences is not displayed for organ pairs where the number of observed cases was no larger than five. The total number of patients is 1009, but given that a single patient does not necessarily have metastases in exactly two organs, the boxes do not add up to this. (b) Differences (in percentages) between the estimated joint probability of a patient having metastases in both corresponding organs (i and j) observed from the actual data and the theoretical values, assuming that metastases appear independently. Metastasis pairs with Arabic numerals indicate that the results remained significant with Bonferroni correction. The rest of the results are nonsignificant or became nonsignificant under multiple testing correction. Lower than expected vs. higher than expected probabilities are highlighted in blue and red, respectively. (c) Order preference of metastatic sites, calculated as the ratio of cases with a metastasis sooner in organ i (x axis) than in j (y axis) out of all cases with metastases in both organs (cases where both metastases were detected at diagnosis or where the date of metastasis appearance is ambiguous were excluded). More common orders are highlighted with yellow boxes, while rare ones with blue. (d) Interarrival times before the appearance of metastases in specific organs. Colored numbers below the box plots display the median results. p values at the upper part of the figure indicate the results of testing (Wilcoxon test) the null hypothesis that the mean elapsed time before the appearance of a metastasis in the given organ is no different from the mean elapsed time before the appearance of a metastasis in any organ. The horizontal dashed line represents the latter value. Early metastases were excluded from the analysis due to the ambiguity in their onset times. (e) Interarrival times after the appearance of metastases in specific organs. p values at the upper part of the figure indicate the results of testing (Wilcoxon test) the null hypothesis that the mean elapsed time after the appearance of a metastasis in the given organ is no different from the mean elapsed time after the appearance of a metastasis in any organ

FIGURE 3

Kaplan–Meier plots for overall survival (OS) in SCLC patients with regards to the primary tumors' localization and the number of metastatic sites. (a) Patients with centrally located primary tumors have decreased survival outcomes compared to those with peripheral tumors, although this tendency did not appear to be statistically significant (p = 0.072). (b–d) Patients with at least one metastatic site at diagnosis exhibited worse median OS than those without metastasis (log‐rank p < 0.0001). Patients with multiple metastases have worse OS than those with no or a single metastasis at diagnosis (log‐rank p < 0.0001). Patients with three or more metastases at diagnosis have lower survival rates than those with at most two metastases (log‐rank p = 0.0027)