Meta-analysis of mutational site-specificity and concordance/discordance characteristics of myeloid sarcoma.

Abstract

6583 Background: Myeloid sarcoma (MS) is an atypical, extramedullary manifestation of acute myeloid leukemia. Next-generation sequencing (NGS) studies have explored mutational profiles of MS tumor sites and bone marrow (BM) samples. However, site-specific molecular patterns and mutational concordance/discordance between MS sites and BM remain unclear. Methods: Relevant studies about MS retrieved from a Pubmed search (1999-2023) with information about MS site involvement and NGS data were included. With R software, a meta-analysis (MA) of descriptive data, MS tumor site, mutations, and mutational concordance/discordance was performed to obtain pooled prevalence estimates using the Freeman-Tukey double arcsine transformation, inverse variance, and random effects model. MA of median age was pooled using the methods of McGrath et al. (2023). Meta-regression (MR) was performed with mixed-effects model. Chi-square test was used for statistical analysis of categorical variables. Results: 87 patients with MS from 10 studies were identified. Median age was 53 years (95% confidence interval [CI]: 48-56), and MS was more common in males (pooled estimate of 67%, 95% CI: 54.8-78.5). Skin was the most common site involved in MS with random effects pooled estimate of prevalence of 39.4% (95% CI: 20.8-59.3), followed by lymph node (10.1%, 95% CI: 1.8-21.8), bone (5%, 95% CI: 0.2-13.3), and soft tissue (2.4%, 95% CI: 0-9.5). Pooled estimates of most common mutated genes were NPM1 (20.3%, 95% CI: 6.2-38), FLT3 (18.3%, 95% CI: 5.5-34.6), TP53 (8.4%, 95% CI: 1.8-17.7), NRAS(6.6%, 95% CI: 0.9-15.4), and IDH2 (4.3% 95% CI: 0-13.4). NPM1 (13/34), FLT3 (7/34), and IDH2 (6/28) were enriched in cutaneous MS. KIT was exclusively enriched in non-cutaneous MS (8/53) compared to cutaneous MS (0/34) (p = 0.017). Although the prevalence of small bowel MS was low (n = 3), FLT3 was mutated in all cases (p < 0.01). Comparing MS tumor site to BM, pooled prevalence of mutation concordance was similar to discordance (49.8% [95% CI: 34.4-65.1] and 50.2% [95% CI: 34.9-65.6], respectively). Most common concordant mutations were FLT3 (5/28) and NRAS (5/28). NPM1 and FLT3 were most common discordant mutations and enriched in MS tumor sites (p < 0.01 and p < 0.05, respectively). NPM1 was a discordant mutation in MS tumor site only. Univariate MR of cutaneous MS found no significant influence from male sex and mutational concordance/discordance. Multivariate MR showed significant interaction with males and mutational discordance toward cutaneous MS (p < 0.01). Conclusions: Our meta-analysis highlights potential site-specific and mutational disparities in MS. Knowledge about such differences may allow the use of targeted therapies against different molecular aberrations, emphasizing the need to sequence both BM and MS sites. Findings suggesting interaction of male sex and mutational discordance in cutaneous MS needs further validation.