Pediatric ALL treatment protocols rely heavily on chemotherapeutic agents such as 6-MP and MTX. Although these drugs have markedly improved survival rates, substantial inter-individual variability in drug response and toxicity remains a significant clinical challenge. Pharmacogenetic markers for thiopurines are already implemented in clinical practice; however, this is not the case for other drugs used in ALL treatments, including MTX. Therefore, there is a clear need for further investigation of pharmacogenetic variants with the potential to personalize ALL therapy. In this study of Croatian pediatric patients with ALL treated at the tertiary care center, we explored associations between germline variants in pharmacogenes and both 6-MP dosing as well as MTX pharmacokinetics and toxicity.

Thiopurine S-methyltransferase (TPMT) is a key enzyme in thiopurine metabolism[19]. and well-established pharmacogenetic marker with CPIC dosing recommendations[13]. Loss-of-function TPMT variants are associated with reduced enzymatic activity and a markedly increased risk of thiopurine-induced myelosuppression.

Unexpectedly, our study did not identify a statistically significant association between TPMT variant and thiopurine-related phenotypes in our cohort. This finding differs from the majority of published studies and suggests that, in the Croatian population, TPMT genotype may have limited predictive value for thiopurine tolerance. This discrepancy may reflect population-specific genetic structure, limited statistical power, or the contribution of additional genetic modifiers and nongenetic factors. Notably, thiopurine intolerance has also been reported in patients with normal TPMT activity, suggesting that TPMT alone may not full explain inter-individual variability. Our finding support the need for population-specific validation of established pharmacogenetic markers and highlight the potential value of multigene approaches.

The ITPA gene, encodes inosine triphosphate pyrophosphohydrolase, which prevents the accumulation of noncanonical and potentially mutagenic nucleotides. Reduced ITPA activity has been associated with increased susceptibility to thiopurine-related toxicity [20, 21].The most extensively studied ITPA variant, rs1127354 has been linked to thiopurine intolerance, including risk of neutropenia and lower tolerated 6-MP doses, although findings vary across populations [22,23,24,25].

In our pediatric ALL Croatian cohort, ITPA rs1127354 heterozygosity was associated with a significantly lower average 6-MP dose compared with wild-type patients (p = 0.045), supporting a reduced-tolerance phenotype consistent with prior reports. This interpretation is consistent with existing meta-analytic evidence where Lee and coworkers [26], reported that ITPA rs1127354 is associated with an increased risk of neutropenia and hepatotoxicity in pediatric patients with ALL, with carriers more frequently requiring dose reductions. Similarly, studies from Asian populations have demonstrated that ITPA rs1127354, either alone or in combination with NUDT15 variants, modulates 6-MP intolerance [27,28,29,30]. We also observed a higher frequency of the variant allele compared with reference European populations, suggesting potentional population specific relevance, while this findings should be interpreted cautiously given the modest sample size. NUDT15 is a key pharmacogene involved in thiopurine metabolism, encoding an enzyme that hydrolyzes cytotoxic thioguanine triphosphates (TGTP/TdGTP) and prevents their incorporation into DNA. Loss of function variants are associated with excessive metabolite accumulation and severe thiopurine-related toxicity [31], and NUDT15 is therefore considered clinically actionable according to CPIC guidelines [32].

In our cohort, the rs61973267 variant in the 3’UTR of NUDT15 was associated with higher 6-MP dose tolerance in heterozygous carriers compared with wild type individuals, although this finding did not reach statistical significance. A similar trend has been reported in a Slovenian pediatric ALL cohort [33], supporting a potential modulatory role of this regulatory variant, although no significant association was observed in two other studies [34, 35].

While most prior studies have focused on coding loss-of-function variants such as rs116855232 [36, 37], our findings suggest that noncoding variants may also contribute to interindividual variability in thiopurine response, potentially through regulatory mechanisms. Although the clinical relevance of rs61973267 remains uncertain, the consistent direction of effect across studies supports further investigation of regulatory NUDT15 variants [12, 38, 39]. These results support current recommendations to include NUDT15 in pharmacogenetic testing, while also suggesting that a broader range of variants may improve prediction of thiopurine tolerance.

Previous studies have linked the PACSIN2 rs2413739 variant to MP-related toxicity during maintenance therapy in pediatric ALL where the TT genotype has been associated with an increased risk of severe GITtoxicity independent of TPMT status [40, 41] and these findings have been replicated across protocols [42], suggesting a potential modifying role of PACSIN2 on thiopurine response through effects on TPMT activity and vesicular trafficking. However, this variant is not currently considered clinically actionable under CPIC guidelines and appears to function as a context-dependent modifier influenced by MP dose intensity and treatment phase.

Pharmacogenetic studies have identified multiple genes contributing to the marked inter-individual variability in MTX response, including genes involved in folate metabolism (e.g., MTHFR, MTR, MTRR, DHFR), and drug transport (e.g., SLCO1B1, SLC19A1, OAT1, OAT3). However, no pharmacogenetic marker is currently considered clinically actionable for MTX dosing according to CPIC guidelines, and genotyping is therefore not routinely recommended.

The SLCO1B1 gene encodes the hepatic organic anion transporter OATP1B1, a key mediator of hepatic MTX uptake and systemic clearance. The rs4149056 variant has been associated with reduced transporter function and altered MTX pharmacokinetics, and is among the extensively studied polymorphisms in this context (43,44,45). In our cohort, we did not observe a significant association between rs4149056 and MTX pharmacokinetic parameters. However, we identified a trend suggesting that patients with the wild-type genotype were more likely to develop oral mucositis compared with variant carriers. This is consistent with findings reported by Liu et al. [6], who observed a reduced risk of mucositis among carriers of the variant allele, although no association was observed in several other studies [44, 46], while at least one study has described an association with mucositis risk [47], highlighting inconsistency across cohorts. Although not statistically significant, this observation is biologically plausible. Reduced OATP1B1 activity in variant carriers may limit hepatic MTX uptake and intracellular polyglutamation, potentially decreasing toxicity in rapidly proliferating tissues such as the oral mucosa.

Taken together, these findings suggest that rs4149056 may act as a nonactionable modifier of MTX toxicity, with clinical relevance influenced by additional genetic and treatment-related factors. Multiple studies have evaluated variants in MTHFR, TYMS, and SLC19A1 in relation to MTX pharmacokinetics and toxicity, with inconsistent results across populations. While some reports associate these variants with altered MTX plasma levels, hepatotoxicity, or GIT, none of these genes are regarded as CPIC-actionable for MTX dosing. In our cohort, we did not observe any significant associations between these variants and MTX pharmacokinetics or MTX-related toxicity, supporting the notion that their individual predictive value is limited.

Taken together, our findings reinforce the distinction between CPIC-defined actionable pharmacogenes and nonactionable modifier genes, emphasizing that variants such as PACSIN2, MTHFR, TYMS, and SLC19A1 may contribute to interindividual variability only in specific clinical contexts and are insufficient as standalone predictors of MTX response or toxicity.

Finally, we evaluated the cumulative genetic contribution to MTX-related multiorgan toxicity and 6-MP dose requirements by constructing a PRS incorporating pharmacogenetically relevant variants. Variants associated with reduced enzymatic activity would be expected to contribute to increased drug intolerance. Although no statistically significant differences in PRS distributions were observed between patients with and without MTX-related multiorgan toxicity, patients maintained on lower average 6-MP doses tended to have higher PRS values. This suggests a possible association between cumulative genetic burden and reduced drug tolerance, although the observed trend did not reach statistical significance (p = 0.08).

Limitations of the present study include the relatively small cohort size (n = 43), which substantially limits statistical power, particularly for the detection of associations involving rare variants. As a result, the study may be underpowered to detect modest genetic effects, increasing the likelihood of both false-negative and false-positive findings. Consequently, the observed associations should be interpreted with caution and considered as hypothesis-generating rather than confirmatory. Furthermore, given the small sample size and low frequency of certain variants (e.g. for TPMT gene), the absence of expected association may reflect limited statistical power rather than true absence of effect. Larger, multicenter studies integrating ITPA into multigene pharmacogenetic models will be essential to clarify its independent clinical utility and to determine whether genotype-guided 6-MP dose optimization improves toxicity management without compromising treatment efficacy. However, the observed interindividual variability in treatment response suggests that a limited candidate-gene approach may not fully capture the complexity of pharmacogenetic influences. This highlights the need for broader analyses, such as next-generation sequencing, incorporating a wider spectrum of pharmacogenetic markers, including genes relevant to other drugs administered concomitantly during ALL therapy. In addition to genetic determinants, variability in MTX pharmacokinetics and 6-MP tolerance is influenced by a range of clinical factors. These include hydration status, renal and hepatic function, infection status, concomitant medications, and the timing of toxicity assessments, all of which can significantly affect drug clearance and toxicity profiles. Owing to the retrospective nature of this study, comprehensive and standardized data on these parameters were not consistently available and therefore could not be systematically included in the analysis. This represents an important limitation, as such factors may confound or modify the observed associations between candidate genetic variants and treatment outcomes, potentially masking true genetic effects. Future prospective studies integrating detailed clinical and pharmacological data alongside genetic profiling are warranted to enable a more accurate assessment of determinants of drug response and toxicity.

The PRS included TPMT rs1142345 and ITPA rs1127354, variants previously implicated in thiopurine metabolism. The absence of statistically significant associations may reflect the multifactorial nature of MTX toxicity and thiopurine intolerance, where genetic effects interact with clinical and treatment-related factors. Given the small sample size and heterogeneity of the study population, these findings should be interpreted with caution.