Secondary Unsupervised Profiling of Monte Carlo–Derived Health-Risk Outputs for Heavy-Metal Exposure From Edible Mushrooms

Abstract

This study presents a secondary, exploratory analysis of scenario-level health-risk outputs derived from the previously published probabilistic risk assessment of Acar et al. (2026). No new mushroom samples, elemental measurements, exposure measurements, or de novo Monte Carlo simulation were generated in the present manuscript. The source study quantified Cd, Pb, As, and Hg in five edible wild mushroom species, namely, Macrolepiota mastoidea, Bovista aestivalis, Tricholoma fracticum, Helvella sp., and Rhizopogon roseolus, and produced probabilistic noncarcinogenic risk outputs for adult and child exposure scenarios. The upstream dataset was based on five field-collected mushroom specimens, each representing one of the evaluated species and reported with a corresponding collection locality and collector number in the source article. In the present analysis, the available Monte Carlo–derived Cd, Pb, and As records were restructured into a secondary matrix containing 1,50,000 Cd-, Pb-, and As-specific scenario records, corresponding to 10,000 Monte Carlo iterations for each metal within each of the five mushroom species. Hg was not included in the multivariate profile model because the available scenario file did not provide a complete Hg-specific risk vector in the same format required for the clustering workflow and because Hg included a nondetect case for Helvella sp. in the source article. Estimated daily intake (EDI), metal-specific target hazard quotient (THQ), and cumulative hazard index (HI/TTHQ) values were standardized and examined using principal component analysis (PCA), uniform manifold approximation and projection (UMAP), and K-means clustering. Comparative clustering for k = 3–6 indicated that k = 4 provided a balanced exploratory solution, although the Silhouette scores (0.36–0.40) and Davies–Bouldin indices (0.83–0.93) indicate modest rather than strong separation. The four profiles showed mean HI values from 3.27 to 15.36 and 95th-percentile HI values from 5.09 to 41.72; therefore, even the lowest relative profile remained above the conventional HI = 1 noncarcinogenic concern threshold. Rhizopogon roseolus was most strongly associated with the higher cumulative-risk/toxic-pressure profile (49.2% of its scenarios), whereas Macrolepiota mastoidea was most frequently assigned to the lower relative cumulative-risk profile (48.0%). These findings should be interpreted as descriptive, data-driven risk profiles rather than causal toxicological classes.