The results suggest that MRI does not predict the consistency of PA through conventional sequences, as well as DWI, ADC, and FIESTA.
Some studies have confirmed the hypothesis that a high percentage of collagen (fibrous tumors) contributes to a low signal on T2 [6, 7, 8, 13]. However, in our study, repeating what was found in other studies, this correlation was not evident [14–19]. Pieralini, unlike Iuchi, found that soft tumors have lower SI than hard or intermediate tumors [6, 13].
The lack of standardization of the MRI analysis is an important limiting factor that may explain the divergence of conclusions, including the techniques for acquiring the sequences and devices with different brands and variable powers (TESLA); it is difficult to measure the influence of these factors on the SI. Other nonmeasurable factors that influence the SI are the bony structures at the base of the skull and degree of aeration of the sinuses, especially the sphenoid. Furthermore, as discussed in the study by Pieralini et al., other factors contribute to the definition of SI on T2, such as fibrous tissue, cellularity, and low nucleus/cytoplasm ratio (low signal); extracellular space would increase free water by raising SI at T2 [6].
However, the amount of fibrous tissue, presence of amyloid, and iron identified in histological study did not show a strong correlation with T2 hyposignal in PAs [15].
In our study, cystic or hemorrhagic components, which alter the SI and influence tumor consistency, did not interfere with the results, as patients with MR with such characteristics were excluded from the study.
In evaluating the DWI/ADC sequences, no correlation was found between SI and tumor consistency, as in other studies that evaluated such a possible correlation [16, 17, 19].
Pieralini et al. found higher ADC values for hard tumors, suggesting a cut-off value of 1 mm2/s to differentiate non-aspirable tumors. Results are not congruent with Suzuki, who found lower ADC values in the intermediate consistency group compared to the soft adenomas group, similar to our study. In our study, the soft tumors had ADC values slightly higher than those of hard tumors but without statistical significance [6, 17].
The analysis of cellularity, nucleus/cytoplasm ratio, extracellular space, and cytoplasmic content contribute to this result. However, such cytoarchitectural characteristics are changeable depending on the resection method used (aspiration, piecemeal). The patient’s own movement during the MRI scan influences the SI on DWI [20].
In evaluating other sequences, such as FIESTA and FIESTA after contrast, there was also no statistical correlation between SI and tumor consistency, despite both showing an increase in SI as tumor consistency increases. The FIESTA sequence was studied by Yamamoto et al., using 3T MRI, in which they found opposite results (SI in CE-FIESTA higher for soft tumors than for hard tumors) [21].
In some studies, the presence of fibrous lesions would be related to the increase in collagen, causing an increase in the consistency of these tumors [11, 13, 22–25]. This fact was not evident in our study, in which the hard tumors did not statistically present a greater amount of collagen, although we found a higher proportion of soft tumors with a small amount of collagen fibers (Table 3). The correlation between tumor consistency and amount of collagen is conflicting. The irregular distribution of collagen in the tumor would be a factor that would make a reliable analysis difficult, especially considering that most PAs are resected with the “piecemeal” technique [14, 26].
In the present study on collagen and MR characteristics, it was not possible to predict the amount of this content through the analysis of the studied MR sequences, corroborating the studies by Suzuki and Mahmoud [17, 19]. While Iuchi demonstrated that PAs with high collagen concentration are hypointense on T2, Pierallini, despite not finding a linear relationship between T2 findings and amount of collagen, showed that a greater amount of collagen leads to a progressive reduction in SI in DWI. Romano et al. found a negative correlation between the SI on post-contrast dynamic T1 sequence and percentage of tumor fibrosis [6, 11, 13].
In an attempt to perform an objective and reproducible collagen analysis, we tested the FD analysis for this function. Until the present study, we did not find in the literature the evaluation of PAs collagen through the analysis of the FD, which is capable of measuring the spatial filling and structural density associated with collagen in the extracellular matrix [26]. After comparing the FD data with the tumor consistency, no correlation was found (Table 2).
Case Example
Male, 65 yo, headache and bitemporal hemianopia.
Labs
hypocortisolism and TSH reduction.
Pre-op MRI (Fig. 1): Size 2.6 x 3.0 x 4.7 cm; Knosp 4; Hardy IVC; IS: ADC 0.8775; DWI 1.0824; T1 1.2545; T1 post-contrast 1.6582; T2 1.2727; Fiesta 2.6153; Fiesta post contrast 1.9696.
Surgery consistency
Soft.
Histopathology (Figs. 2 and 3): Ki67: 1%; FSH positive immunohistochemistry; Masson 1,715; Reticulin 1,678.
Study Limitations
We recognize some limitations of our study, among them are the limited sample size of patients due to the difficulty of establishing standardized examination protocols and the research design in public health services in our country.