To the best of our knowledge, this is the first study to investigate in subjects of both sexes whether PRL values, once any increase due to stress is correctly excluded, can predict the presence of a pituitary disease. Specifically, our data show that after excluding PRL values that are already diagnostic for pituitary tumor-related etiology, PRL levels can predict with modest accuracy the presence of a pituitary lesion, but cannot predict a macrolesion.

Certainly, despite guidelines [5, 12] and various authors [13, 15] suggest always performing a pituitary imaging, even in the case of mild hyperPRL, this is not always applicable in real practice for several reasons, including economic and availability ones.

If on the one hand, however, the failure to diagnose a possible microlesion generally does not raise particular concern, except in rare cases characterized by hormonal hypersecretion of another nature, starting an ex adiuvantibus treatment with dopaminergic agonists may potentially expose patients to relevant side effects [18]. Furthermore, in certain cases this may lead to delayed detection of pituitary macrolesions which may later manifest with mass effects, resulting in the deterioration of long-term outcomes. Consequently, the identification of a PRL threshold value beyond which a MRI must necessarily be performed could allow resources to be optimized, from a cost-benefit perspective.

In 1996 it was proposed by Rand et al. [10] to use a PRL threshold >100 µg/L as a cut-off for considering imaging of the pituitary region: it is clear, however, that this value cannot be considered reliable in our series since it would have led to the loss of 74 out of 106 (69.8%) sellar masses and in particular of 12 out of 27 (44.4%) macrolesions.

Undoubtedly, we have to consider that in line with prior studies [10, 13, 14], we have identified a considerable prevalence of pituitary abnormalities in patients diagnosed with hyperPRL.

In their study, on the other hand, Souter et al. [15] observed a significantly lower prevalence of pituitary lesions with 60.9% of MRIs being negative. A distinctive aspect of this study was its exclusive focus on patients with mild-to-moderate hyperPRL (i.e. <100 µg/L) on a single sampling, although repeated on a second occasion. Consequently, it exists the possibility that a proportion of the subjects considered by such study were not “truly” hyperprolactinemic and that the reported levels could have turned normal upon serial sampling [6]. In fact, it has been shown that the highest effectiveness of serial PRL assessment occurs precisely for those patients with borderline values, while levels >94 µg/L showed 97% specificity in correctly discriminating patients with true hyperPRL [6]. In line with this, even considering only the subgroup of patients with mild-to-moderate hyperPRL in our population, the number of patients with negative MRI would still have been no more than 36% in females and not even 20% considering the male population.

As a result, patients with hyperPRL at serial sampling exhibit a notable pre-test probability of harboring a pituitary mass and a high PRL value, especially beyond 100 µg/L, as previously reported [10], is strongly indicative of an underlying pituitary cause (100% of cases in our study), once secondary causes are excluded.

On the other hand, lower PRL levels do not completely exclude the possibility of a pituitary mass, even a voluminous one, because of the aforementioned stalk effect [3]. In this context, therefore, PRL values seem more useful as a rule-in rather than a rule-out test and in light of all this, it seems to be safer to perform an MRI of the sellar region in all cases of confirmed hyperPRL, similar to what had already been suggested [13, 15]. Of note, no differences regarding the age of patients with and without a pituitary abnormality were observed, either considering the entire cohort or stratifying by sex.

Furthermore, in contrast to previous studies [13, 14], our statistical analysis excluded patients with both macroPRL, who should not undergo further diagnostic testing, as well as subjects with PRL levels indicative of a pituitary etiology ( >250 µg/L and >500 µg/L) [5, 8]. After such patients were excluded from the analysis no correlation between PRL values and greatest lesion diameter was observed, contrary to previous reports in the literature [13, 19]; however, significance was readily observed once these patients were included again in the analysis. This may be explained by the fact that our analysis was not limited to patients with a definite diagnosis of prolactinoma and that a likely large number of non-secreting lesions were present in our cohort.

Finally, our study also evaluated a significant proportion of male subjects and seems to confirm that in males idiopathic hyperPRL is not a common condition [20] and therefore always worthy of further diagnostic investigation, even for very modest elevations especially in case of symptoms such as ED.

Compared to females, men diagnosed with hyperPRL were significantly older in our cohort; moreover, lesion size seems to be typically larger [21], although a statistically significant difference was not achieved in our case.

In relation to the female cohort, on the other hand, it is noteworthy to highlight that 80% of women with a concomitant diagnosis of PCOS demonstrated an underlying pituitary lesion. Despite the limitations of our small sample size, these findings provide support for the idea that hyperPRL is not an intrinsic component of PCOS and that should not be underestimated [22].

Our study presents some strengths and limitations. One notable strength of our study is the rigorous exclusion of hyperPRL through serial sampling. Consequently, we can confidently exclude the inclusion of hyperPRL cases that were secondary to venipuncture stress among the patients analyzed. Another strength was that macroPRL was excluded in all cases. In addition, our population is larger than the aforementioned studies [13,14,15] and also included a large component of male subjects. Lastly, our patient cohort comprised individuals referred to our medical attention for various reasons, not solely limited to issues concerning couple infertility (just 10% of subjects). As a result, it is likely that our study population is representative of the typical clinical practice encountered by most endocrinologists.

Limitations of the study include its retrospective nature and the fact that MRIs of the sellar region were not all performed at the same center.