Association of mediastinal lymphadenopathy with COVID-19 prognosis

Xavier Valette and colleagues¹ reported a high (66%) prevalence of mediastinal lymphadenopathy in 15 patients with COVID-19 admitted to their intensive care unit (ICU), an approximately 11-fold discrepancy with systematic reviews reporting pooled prevalence of 3·4%² and 5·4%.³ This topic deserves further investigation, especially considering that small sample sizes imply large confidence intervals.

We retrospectively reviewed 410 patients with COVID-19 (including 288 male and 122 female patients; median age of all patients 68 years [IQR 57–78]) who underwent CT at emergency department admission in three hospitals in Lombardy, Italy (Fondazione Poliambulanza Istituto Ospedaliero, Brescia; ASST Crema, Ospedale Maggiore, Crema; ASST Santi Paolo e Carlo, Ospedale San Paolo, Milan), from Feb 21 to March 18, 2020, during the pandemic peak in Lombardy. 76 patients had mediastinal lymphadenopathies (ie, lymph nodes with a short-axis diameter >1 cm), giving a prevalence of 19% (95% CI 15–22).

Whereas our CT examinations were done at emergency department admission, Valette and colleagues' data¹ derive from patients in the ICU. Thus, our lower lymphadenopathy prevalence could be explained by the lower severity illness of our patients. However, 60 (15%) patients in our cohort were admitted to the ICU, of whom only 15 (25%, 95% CI 14–36) had lymphadenopathies at emergency department admission (appendix).

Valette and colleagues¹ hypothesised that disease severity could probably explain the discrepancy between previous data and their ICU population. After applying the Bonferroni correction for multiple comparisons to our series of patients (obtaining a p value threshold of 0·003, above which p values were not significant), we found no significant differences between patients with and without lymphadenopathies in terms of sex, age, history of cancer, non-invasive ventilation or ICU admission during hospitalisation, length of hospital stay, laboratory findings, and CT features such as parenchymal involvement and disease progression, both assessed according to the classification by Bernheim and colleagues⁴ (appendix). However, lymphadenopathies at admission were significantly more frequent in patients with a crazy paving pattern on CT than in those without (33 [31%] of 106 vs 43 [14%] of 304, p<0·001) and in patients who died during hospitalisation than in those who were discharged (37 [27%] of 136 vs 39 [14%] of 274, p=0·001; appendix).

Although invasive microbiological samples were not available for our patients (so we cannot exclude bacterial or fungal coinfections), our lymphadenopathy prevalence was lower than that reported by Valette and colleagues¹ but three times higher than estimates for other populations.2, 3, 5 We therefore agree in defining lymphadenopathy as a “not-atypical” feature of COVID-19. Furthermore, our data suggest that lymphadenopathy may be considered a predictor of a worse outcome. The pathophysiological meaning of this finding in relation to host response to virus infection and the possibility to use this information in the clinical management of patients with COVID-19 remain to be investigated.