This Brazilian multicenter cohort provides the largest Latin American dataset to date on the use of TMZ as a single therapy in aggressive and metastatic pituitary adenomas. The radiological disease control rate of 93.3%, including 23.3% partial response and 70% stable disease, supports the activity of TMZ in this setting. While no complete responses were observed, this finding is consistent with the broader literature, in which complete radiological responses are uncommon. Recent meta-analyses have reported partial response and stable disease as the most frequent radiological outcomes (32–33% each), whereas complete response is rare (0.6–4%), and progressive disease remains common (29%) [2, 26]. However, direct cross-study comparisons should be interpreted cautiously because retrospective cohorts differ in patient and tumor characteristics, prior treatments, and response assessment timing, and standardized follow-up protocols are lacking.

Compared with previously published studies, our cohort from the Brazilian multicenter study demonstrated distinct demographic characteristics. With a mean age of 29.5 years, the Brazilian participants were notably younger than those in other cohorts [13, 18, 27, 28], whose mean ages ranged from 40.5 to 58.3 years. Additionally, a greater proportion of women was observed in the Brazilian cohort (53%), whereas the other studies predominantly included male participants, with rates reaching up to 87.5% [13, 18, 27, 28]. These findings suggest potential differences in the epidemiological profile or inclusion criteria across studies, which should be taken into account when interpreting and comparing the results.

Importantly, our results offer novel insights when patients are stratified by tumor subtype, size, functionality, and aggressiveness. Patients with acromegaly exhibit a notably high rate of partial radiological response and no disease progression, a finding that may reflect a distinct somatotroph tumor biology or increased sensitivity to alkylating agents. In contrast, patients with prolactinomas and Cushing’s disease experienced disease progression, suggesting possible lineage-related differences in TMZ responsiveness, which merits further investigation. Within the corticotroph spectrum, Crooke cell pituitary adenomas are a high-risk variants that are often associated with aggressive behavior [29, 30]. The stable disease observed in the Crooke cell population in our cohort suggests that even histologically aggressive subtypes may retain clinically relevant chemosensitivity to temozolomide. Metastatic pituitary adenomas exhibited a dichotomous pattern. Although the small sample precludes definitive conclusions, the data highlight the therapeutic challenge posed by metastatic pituitary adenomas and the need for alternative strategies in this subgroup.

The unpredictable clinical behavior of APT is partly explained by the absence of robust prognostic markers at diagnosis. In our series, Ki-67 and p53 expression alone did not reach statistical significance in predicting the tumor response, corroborating previous evidence that conventional proliferation markers may be insufficient when evaluated independently [31, 32].

Although MGMT expression and promoter methylation have been proposed as predictors of TMZ responsiveness, their utility remains inconsistent and was not assessed in this cohort [22, 33, 34]. We did not assess MGMT expression in our series, which represents a limitation given the role of this DNA repair enzyme in counteracting TMZ-induced cytotoxicity. However, the predictive value of the MGMT status remains inconsistent across cohorts, and the recent ESE consensus no longer recommends routine MGMT immunohistochemistry prior to a trial of TMZ, mainly owing to technical variability, temporal changes in expression, and the lack of alternative therapeutic options in this setting [2]. Thus, while MGMT remains of scientific interest, its clinical utility is still limited, underscoring the need to identify more reliable biomarkers of TMZ response and resistance in APT.

The biochemical responses of functioning tumors are heterogeneous. Although limited by sample size, radiological and hormonal responses may not always correlate, and endocrine control should be independently monitored in clinical practice. A hormonal response, even in the absence of a complete radiological response, may hold substantial clinical significance by contributing to morbidity reduction. The biochemical response rates to TMZ in functioning tumors range widely from 19% to 100% [13, 18, 20, 21, 28, 35,36,37,38,39,40]. In our cohort, partial biochemical response was most common in prolactinomas and acromegaly, whereas Cushing’s disease displayed the broadest spectrum of biochemical response. Taken together, our results underscore the need for harmonized definitions of biochemical response to enable comparability across studies. Consistent with previous observations, biochemical control in functioning pituitary adenomas was variable and did not always parallel the radiological response. Notably, a recent single-center study identified early initiation of TMZ and low MGMT expression as independent predictors of favorable outcomes, particularly in functional tumors [41]. Nevertheless, many studies have traditionally focused on radiological outcomes as the primary endpoint, relegating hormonal response to a secondary outcome [2, 13, 35]. The literature shows considerable variability in the definition of hormonal response. While some studies define response as a reduction exceeding 50% in circulating hormone levels [13, 34], others adopt a stratification approach analogous to the RECIST criteria, since biochemical response is defined as complete normalization of hormone levels; partial response corresponds to a reduction exceeding 20%, stable disease is defined as a fluctuation within ± 20%, and progressive disease is defined as an increase of more than 20% in hormone levels [21, 42]. We used the latter approach for our evaluations.

An extensive survey demonstrated that clinically functioning tumors were more likely to respond to TMZ than nonfunctioning tumors were, independent of the MGMT status [21]. In our series, however, our radiological response rates were similar between functioning and nonfunctioning tumors, with both partial response (23.8% vs. 22.2%, respectively) and radiological disease control, including partial response or stabilization (90.4% vs. 100%). Therefore, functional status alone may not consistently discriminate responsiveness in real-world practice. This apparent discrepancy could reflect differences in patient selection, the timing of TMZ initiation, or referral bias across cohorts. In addition to these differences, one criterion for the definition of aggressiveness in pituitary tumors relies on resistance to standard therapies, which is more difficult to apply to nonfunctioning tumors. In functioning tumors, medical therapies such as cabergoline or somatostatin analogs provide a clear benchmark of resistance, whereas in nonfunctioning tumors, standard therapy is essentially surgical and adjuvant radiotherapy, with limited medical alternatives. Although cabergoline has been investigated in nonfunctioning pituitary tumors, with small series reporting disease stabilization and modest tumor shrinkage, it has not been systematically evaluated in a subset of aggressive nonfunctioning pituitary adenomas [43,44,45,46]. These conceptual differences underscore the challenges in establishing uniform criteria for aggressiveness across different tumor subtypes. Importantly, while functioning tumors allow for the additional endpoint of biochemical control, our data reinforce that radiological assessment remains the critical measure of efficacy across all subtypes.

Radiological and biochemical endpoints are not fully interchangeable in functioning tumors. Although overall concordance was high, a clinically relevant discordance emerged: biochemical progression occurred in 12.5% of patients despite radiological disease control. This finding supports independent monitoring of imaging and biochemical markers during TMZ, particularly in functioning pituitary adenomas where biochemical changes may directly translate into morbidity. The moderate kappa coefficient observed despite high percent agreement likely reflects the highly unbalanced distribution of radiological outcomes (predominance of PR/SD and rare radiological PD).

Padovan et al., in a systematic review focused on temozolomide (TMZ) discontinuation, reported progression after TMZ withdrawal in 13/52 patients (25%) and emphasized the absence of prospective data to define the optimal treatment duration; they also noted that, in the absence of prospective studies, most investigators tended to extend TMZ therapy to 12 months when effective. In the same discussion, they stated that there is no international consensus on treatment choice at recurrence after an effective TMZ course, although a second TMZ challenge seems reasonable, and that the reported response rate to rechallenge is more or less halved in the literature. Consistent with this clinical uncertainty, loss of disease control after TMZ discontinuation was frequent in our cohort, including among patients who achieved SD or PR while on treatment. TMZ rechallenge provided temporary disease control in some patients (SD/PR), but responses were not uniformly durable, and progression still occurred in a subset. These findings may be clinically relevant when considering treatment duration and postdiscontinuation surveillance, but they should be interpreted cautiously given the small sample size, retrospective design, and clinical heterogeneity.

The safety profile of TMZ in our cohort was globally acceptable, with adverse events documented in 66% of patients, predominantly grade 1. Nausea and myelotoxicity were the most frequent toxicities, and treatment discontinuation was required in four patients (13%) because of grade 3–4 events, with no deaths related to toxicity. These findings align with the broader literature synthesized in the recent international consensus on APT and metastatic pituitary adenomas [2, 13, 47]. Discontinuation rates due to adverse events range from 6% to 15%, most often driven by pervasive fatigue, nausea, or cytopenias, while rare but severe complications such as aplastic anemia, hepatotoxicity, or secondary hematologic malignancies have been reported [2, 13, 47]. Compared with these data, our series underscores two complementary perspectives: first, the burden of mild adverse effects is common and expected in real-world settings; second, that clinically significant hematological toxicity requiring drug withdrawal is not negligible and should be anticipated in routine practice. Importantly, while we did not observe the rare life-threatening complications highlighted in the consensus, our findings reinforce the need for vigilant hematologic monitoring, as hematologic toxicity remains the leading barrier to treatment continuity in this patient population.

Our study revealed significant heterogeneity in the duration of TMZ treatment, reflecting the current lack of well-defined protocols for the management of APT and metastatic pituitary adenomas. The mean duration of TMZ use was 10.7 months (range, 3–44 months), highlighting the variability in clinical decisions to initiate TMZ. In some cases, treatment was discontinued after six months due to disease stabilization, emphasizing the need for protocols that support both earlier initiation and longer treatment courses. Given that TMZ is most often prescribed following documented disease progression, tumor stabilization should be considered a favorable therapeutic outcome.

According to the 2025 ESE Clinical Practice Guidelines [2], for patients who respond to a first course of TMZ—defined as either partial tumor regression or stabilization after rapid progression in the preceding six months—it is now recommended that therapy be continued for at least 12 months and further guided by efficacy and tolerability. Extending treatment beyond 24 months should be carefully weighed against the potential for cumulative severe toxicity. In line with these recommendations, an Italian real-life study including patients treated for more than 12 continuous cycles reported that prolonged TMZ administration was safe and associated with durable disease control and hormonal improvement, particularly among responsive functioning pituitary adenomas [28].

In our study, 10 patients received TMZ treatment for 12 months or longer. Among them, 81.8% experienced at least one adverse event, most of which was mild in severity. Three cases of myelotoxicity were observed, leading to treatment discontinuation in two patients. The treatment discontinuation rate due to adverse effects was higher in the group that used TMZ for more than 12 months (30% vs. 5%), whereas the myelotoxicity rate was similar between the groups (30% vs. 30%). This suggests that cumulative toxicity and treatment fatigue, rather than hematologic events per se, may limit the long-term tolerability of TMZ in APT.

In the present cohort, despite a tumor control rate of 93.3% with TMZ, mortality was observed in 7/30 patients (23.3%), two due to direct tumor-related complications (mass effect) and five due to disease progression after TMZ withdrawal, despite having initially achieved stable disease (n = 3) or partial response (n = 2). These findings reinforce that, in the absence of severe adverse effects, maintaining TMZ in patients with a favorable response may be a safe and effective strategy, even when complete radiological remission is not achieved.

This retrospective multicenter cohort reflects real-world practice in Brazilian tertiary referral centers, where temozolomide is typically reserved for highly pretreated aggressive or metastatic pituitary adenomas. Although prespecified response definitions were applied, heterogeneity in treatment timing, imaging schedules, and biochemical assays across centers, along with missingness in key pathology variables (Ki-67/p53) and limited paired biochemical data, may introduce measurement variability and constrain subgroup and concordance analyses. Therefore, the results should be interpreted as descriptive and hypothesis-generating, and extrapolation to earlier-line use or to settings with different monitoring practices should be made cautiously.

Systematic germline or somatic genetic testing was not performed in this retrospective multicenter cohort. Because genetic testing was not uniformly available across participating centers, it was not included as a predefined variable. We acknowledge this as a limitation, particularly for pediatric or young-onset cases, in which an underlying genetic predisposition may be clinically relevant.