Tumor T2 Signal Intensity and Stalk Angulation Correlates with Endocrine Status in Pituitary Adenoma Patients: A Quantitative 7 Tesla MRI Study

Abstract

Purpose:

Pituitary adenomas are common CNS tumors that can cause endocrine dysfunction due to hormone oversecretion and by mass effect on the normal gland. The study of pituitary adenomas and adjacent sellar anatomy with high-resolution 7T MRI may further characterize endocrine dysfunction. The purpose of this study was to determine the efficacy of 7T MRI in identifying radiological markers for endocrine function.

Methods:

MR images obtained in 23 patients with pituitary adenomas were reviewed by consensus between three neuroradiologists. Landmarks and criteria were devised to measure radiological features of stalk, tumor, and normal gland. Fischer’s exact tests and nominal logistic regression were performed.

Results:

Mean cross-sectional area of the stalk just below the infundibular recess was 6.3±3.7mm². Mean curvature and deviation angles were 34.2°±23.2° and 29.7°±17.3°, respectively. Knosp scores obtained differed between 7T and lower field strength scans (P<0.0001 [right] and P=0.0006 [left]). Ability to characterize tumor was rated higher at 7T compared with lower field MRI, P=0.05. Confidence in visualizing normal gland was also higher using 7T MRI, P=0.036.

The six hormone-secreting tumors had higher corrected T2 mean SI than non-secreting tumors (2.54 vs. −0.38, P=0.0196). Seven patients had preoperative hypopituitarism and had significantly greater stalk curvature angles than patients without hypopituitarism (71.7° vs. 36.55°, P=0.027).

Conclusion:

Radiological characterization of pituitary adenomas and adjacent native pituitary tissue may benefit with the use of 7T MRI. Corrected T2 SI of tumor may be a sensitive predictor of hormonal secretion and may be useful in the diagnostic work-up for pituitary adenoma.

Summary statement:

7T MRI may be valuable in identifying markers of endocrine function in patients with pituitary adenomas. Our results indicate that hormone-secreting tumors have higher T2-weighted SI and tumors associated with preoperative hypopituitarism have greater stalk curvature angles.

1. Introduction

The pituitary stalk and gland are anatomically complex structures that are critical in hormonal regulation. Anterior and posterior pituitary lobes receive input from the hypothalamus in distinct ways. While the neurohypophysis is innervated directly by neurons of the hypothalamo-hypophyseal tract, communication of hypothalamus with the adenohypophysis is subserved indirectly by the venous hypophyseal portal system.[1–3]

Pituitary adenomas are common skull base tumors that may cause symptoms through hormonal secretion or, when large, may be symptomatic from compression of the stalk and/or gland. Various radiological features of pituitary adenomas and accompanying hormonal changes have been reported.[4–9] However, earlier studies have employed either 1.5T or 3T MRI, and were thus relatively limited by low signal-to-noise ratio (SNR). Investigations of the pituitary stalk have been similarly limited by the structure’s small size, and few studies report quantitative properties of the normal infundibulum.[1,10,11] Furthermore, quantitative analysis of the stalk becomes increasingly difficult with large neoplasms that cause mass effect, and high-resolution imaging techniques are critical for adequate visualization.

High-resolution quantitative imaging of the pituitary stalk, tumor, and normal gland in the context of pituitary adenomas may further characterize effects of sellar neoplasms on pituitary anatomy, and may hold prognostic value for endocrine abnormalities in adenoma patients. Stalk features such as angulation[9], cross-sectional diameter[12], and signal intensity (SI)[13] have been implicated in endocrine dysfunction by pituitary adenomas, however, precise imaging metrics may help evaluate their contributions to hormone abnormality in adenoma patients. The present study employs 7T MRI to image the sellar and parasellar regions in 23 patients with pituitary adenomas to characterize potential radiological markers of endocrine functional status in these patients.

2. Methods

2.1. Participant Selection

Twenty-five patients with pituitary adenomas were recruited through their neurosurgeon (RS) for this prospective study. Patients were eligible for this study if they were diagnosed with pituitary adenoma, had not received prior surgery or radiotherapeutic intervention before this study, and received neurosurgical resection after imaging. Approval was obtained from the Mount Sinai Medical Center Institutional Review Board prior to initiation. All patients provided written informed consent. Patients were enrolled and scanned at 7-Tesla MRI between September 2014 and February 2019 at the Mount Sinai Medical Center. All patients presented with pituitary adenoma without prior surgical or radiotherapeutic treatment.

2.2. Imaging Protocol

The Institutional Review Board-approved protocol employed a 7T whole-body MRI scanner (Magnetom, Siemens Healthcare, Erlangen, Germany) with a SC72CD gradient coil (G_max=70 mT/m, max slew rate=200T/m/s), and used a single channel transmit and 32-channel receive head coil (Nova Medical, Wilmington, MA, USA). Sequences included a T₁-weighted MP2RAGE[14] sequence: TE(ms)=3.62, TR(ms)=6000, TI(ms)=1050/3000, flip angle (FA)=5°/4°, field of view (FOV)=224×168-mm², slices=240, resolution=0.7mm³ isotropic, scan time=8:08mins. Quantitative T₁-maps were derived from the MP2RAGE sequence. Coronal-oblique and axial T₂-weighted turbo spin echo (T₂-TSE) (TE=60ms, TR=6000ms, FA=180°, FOV=200×168mm², in-plane resolution 0.4×0.4-mm², slice thickness=2mm, slices=60, time=6:50mins) sequences were acquired. A T₁-weighted MPRAGE sequence was also obtained (TE=4.1ms, TR=3000ms, TI=1050ms, FA=7°, resolution =0.7mm³ isotropic, scan time (min)=7:40). All patients had a preoperative diagnostic MRI completed at clinical field strength (1.5T or 3T) using a routine pituitary protocol.

2.3. Morphometric Analysis of the Pituitary Gland and Infundibulum

In vivo measurements and ratings were generated by consensus of three CAQ-neuroradiologists with 3, 13, and 20 years of experience. Tumor dimensions were measured in the transverse, anterior/posterior, and craniocaudal dimensions, and volumes were calculated using a validated ellipsoid formula[15]:

$$\mathit{\text{Volume}}=\frac{\pi }{6}\times \left(AP\times SI\times CC\right)$$

Lesion location was identified as left, right, or central. Cross-sectional area of the infundibulum was measured in the plane perpendicular to the infundibulum just below the infundibular recess on the T1-map (Figure 1). SI of the infundibulum was calculated at the same level but with smaller region of interest (ROI) to mitigate partial volume averaging of suprasellar CSF. To normalize SI values of the infundibulum, SI of the corticospinal tract within each peduncle and of bilateral amygdalas were recorded as reference white and gray matter values, respectively. SI of the infundibulum, corticospinal tracts, and amygdala were also measured on axial T2-TSE imaging at the same levels (Figure 1). The following equation was used to normalize SI values for mean T1 and T2 SI and SI heterogeneity:

$$SI\left(\mathit{\text{Corrected}}\right)=\frac{SI\left(\mathit{\text{Raw}}\right)-SI\left(\mathit{\text{Average GM}}\right)}{SI\left(\mathit{\text{Average WM}}\right)-SI\left(\mathit{\text{Average GM}}\right)}$$

Figure 1.

T1-map SI measurements of normal white (corticospinal tracts in the middle third of the peduncles) and gray matter (amygdala) taken on the same slice as the pituitary stalk SI measurement (A). CSF appears hyperintense on this T1-map is inverted. Cross-sectional area measurement of the pituitary stalk (C) on axial T1-map, taken just below the infundibular recess (white arrowhead, B). SI of the pituitary adenoma (black arrow), normal gland (black arrowhead), healthy white matter (anterior basis pontis, white arrow), and healthy gray matter (medial temporal cortex, white arrowheads), measured on axial T2-TSE images (D).

Two angles of the infundibulum and stalk were measured using the T1-map. The infundibular deviation angle reflects the maximum leftward or rightward deviation of the infundibulum from midline. For this calculation, the midline was defined superiorly by the insertion of the falx cerebri into the superior sagittal sinus, and inferiorly by the midpoint between the inferomedial borders of the inferior foramina ovalia. The curvature angle of the stalk, measured on the coronal oblique plane that included the maximum representation of the stalk, was defined as the difference in angulation between upper and lower tangential planes representing the extremes of local stalk angulation. This measure was subtracted from 180° to obtain final curvature angle. The measurement plane was kept just thin enough to ensure representation of the stalk in a single image. Deviation and curvature angle measurements are shown in Figure 2.

Figure 2.

Diagrammatic rendering of the sella turcica and surrounding anatomy in the presence of a pituitary macroadenoma (A). Vertical blue line represents midline, defined superiorly as the insertion of the falx cerebri into superior sagittal sinus and inferior by the midpoint between the right and left formina ovalia. Infundibular deviation angle reflects deviation of the stalk from this midline (B). Pituitary stalk curvature angle (C) is defined by the angle between the extremes in angulation of the stalk.

Signal intensities and heterogeneities of lesion and healthy pituitary were also quantified using multiple sequences. Lesion measurements were taken using the largest possible ROI excluding any cystic or hemorrhagic component. SI measurements were not taken for lesions more than 95% cystic. Heterogeneities of lesions and healthy glands were quantified by calculating the standard deviation in SI in the ROI. Mean SI and heterogeneity were measured for the lesions and healthy glands using axial T2-TSE. Normalization measurements were collected at the medial temporal gray and white matter at the level corresponding with the ROIs. Mean SI and heterogeneity of tumors, healthy glands, and gray and white matter normalization regions were also measured on T1w-MPRAGE images.

2.4. Lesion Feature Comparison Between Field Strength

Healthy gland and lesion conspicuity were compared at both 7T and clinical field strengths (either 1.5T or 3T). By consensus, neuroradiologists determined which field strength was better for visualizing healthy gland, and the confidence that normal gland was visible at 7T and clinical field strength. For the purpose of this comparison, the sequence that provided the best visualization of the gland was used. All images used for comparison were non-contrast. Confidence that the normal gland was seen was rated in the following manner: 0=not seen, 1=possible, 2=probable, 3=definite. 7T was also compared with clinical field strength for ability to visualize lesion presence as well as lesion characterization. Lastly, Knosp scores were calculated on both 7T and clinical field strength MRI (a and b subscores denote contact with dura).

2.5. Endocrine Analysis

Hormone levels including adrenocorticotropic hormone (ACTH), cortisol, prolactin, insulin-like growth factor-1 (IGF-1), growth hormone (GH), thyroid stimulating hormone (TSH), thyroxine (T4), LH, FSH, testosterone and estradiol were collected pre and post-operatively from patients via phlebotomy when clinically indicated. Tumors were classified as either secreting or non-secreting based on histological findings from surgical specimens. Preoperative hypopituitarism was defined as biochemical undersecretion of one more pituitary hormones or requirement of hormone replacement therapy.

2.6. Statistical Analysis

Statistical database software (JMP version Pro 14) was used for statistical analysis. Variables were assessed for normality using the Shapiro-Wilk test. The Fisher’s exact test was used to detect statistical differences in categorical variables. The Wilcoxon method was used to detect statistical differences in nonparametric continuous variables. Linear regression analysis used for association between imaging and endocrine variables. Nominal logistic regression used for categorical variables. Receiver operating characteristic (ROC) curve analyses were performed for sensitivity and specificity and area under the curve (AUC). P≤0.05 was used as threshold for significance.

3. Results

Two patients did not proceed to immediate neurosurgical resection and a total of 23 patients met inclusion criteria for this study. Most patients (19/23, 82.6%) presented with a clinical scan enabling comparative review, including 9 (47.4%) at 1.5T and 10 (52.6%) at 3T.

3.1. Comparison between 7T and clinical field MRI

From the 23 patients scanned at 7T, 19 (82.6%) had a clinical scan that enabled comparative review, including 9 (9/19, 47.4%) at 1.5T and 10 (10/19, 52.6%) at 3T. Knosp scores measured at 7T significantly differed from scores obtained with lower field MRI on both he left (P=0.0006) and right sides (P<0.0001) (Table 1). An example of discordant Knosp scores between 3T and 7T is depicted in Figure 3.

Table 1.

Left (P = 0.0006) and right (P < 0.0001) Knosp scores measured at 7T and lower field strength. Bolded values indicate concordance in Knosp score between 7T and low field MRI.

		7T Left
		0	1	2	3a	3b	4a	4b
Low field Left	0	7	-	-	-	-	-	-
	1	2	1	1	-	-	-	-
	2	1	-	1	-	-	1	-
	3a	-	-	-	1	-	-	-
	3b	-	-	-	-	2	-	-
	4a	-	-	-	-	-	1	1
		7T Right
		0	1	2	3a	3b	4a	4b
Low field Right	0	9	1	-	-	-	-	-
	1	1	3	-	-	-	-	-
	2	-	-	2	-	-	-	-
	3a	-	-	-	-	-	-	-
	3b	-	-	-	-	1	-	-
	4a	-	-	1	1	-	-	-

Figure 3.

Example of a pituitary adenoma with a different Knosp score at 3T (left images) compared with 7T (right images) on T2-TSE imaging. At 3T, the left and right Knosp scores were 1 and 3b, respectively. At 7T, the left and right Knosp scores were 0 and 3b, respectively. White arrowheads in the coronal (upper right) and axial (lower right) 7T images denote the signal hypointensity confirming normal gland, a distinction that at 3T could not be resolved on axial imaging and was equivocal on coronal scans.

In overall lesion detectability, 7T was rated superior in three patients (3/19, 15.8%), including against two 1.5T patients (2/9, 22.2%) and one 3T patient (1/10, 10%). A clinical field strength scan (1.5T) was judged better in one patient (1/19, 5.3%). There was no difference in conspicuity between field strengths in 15 patients (15/19, 78.9%).

In lesion characterization, 7T was judged better overall in 12 patients (12/19, 63.2%, P=0.0573) compared with clinical strength scans (including 8 patients at 1.5T and 4 patients at 3T), whereas the clinical field strength was better in two (2/19, 10.5%) including one 1.5T patient and one 3T patient (Figure 4). There was no difference in lesion characterization for the remaining five (5/19, 26.3%) patients.

Figure 4.

Example of a pituitary adenoma with improved lesion characterization at 7T (right image), compared with 3T (left image) on T2-TSE imaging. Imaging suggests a cystic tumor with a right anterior tumor nodule, with differing distinction between tumor and normal gland. At 7T, a curvilinear hypointense band (designated by 2 arrowheads) separates tumor medially from normal gland laterally. No such boundary is resolved at 3T so the healthy gland cannot be differentiated from tumor.

In depiction of normal gland, 7T MRI was better than 1.5T in 7 patients (7/9, 77.8%) and equivalent in 2 patients (2/9, 22.2%). 7T depicted normal gland better than 3T in 6 patients (6/10, 60%), whereas 3T was superior in 3 patients (3/10, 30%). There was no difference between 7T and 3T in 1 patient (1/10, 10%). The average confidence rating for visualization of normal gland at 7T (2.27) was significantly greater than the rating at lower field strength (1.63), P=0.036. Normal gland depiction comparison at 3T and 7T is shown in Figure 4.

3.2. Radiological Features of the Infundibulum, Tumor, and Normal Gland

Mean cross-sectional area of the infundibulum immediately below the infundibular recess was 6.3 mm² (σ=3.7 mm²). Mean corrected T1W and T2W SIs of the infundibulum at this level were −0.11 (σ=0.94) and −0.04 (σ=1.1), respectively. Mean deviation angle and curvature angle of the infundibulum were 29.7° (σ=17.3°) and 34.2° (σ=23.2°), respectively. The infundibulum was deviated to the left in 7 cases (7/23, 30.4%) and to the right in 13 cases (13/23, 56.5%) cases. No infundibular deviation was measured in 3 cases (3/23, 13.1%) (Table 2).

Table 2.

Patient demographical and radiological measurements of the pituitary infundibulum, lesion, and healthy gland at 7 Tesla MRI.

Demographics	Value (N = 23)
Male, N (%)	11 (47.8)
Age, μ (σ)	46 (13.7)
Infundibulum
Cross-sectional area (mm2), μ (σ)	6.3 (3.7)
Corrected T1w signal intensity, μ (σ)	−0.11 (0.94)
Corrected T2w signal intensity, μ (σ)	−0.04 (1.1)
Curvature angle (°), μ (σ)	34.2 (23.2)
Deviation angle (°), μ (σ)	29.7 (17.3)
Left deviation, N (%)	7 (30.4)
Right deviation, N (%)	13 (56.5)
No deviation, N (%)	3 (13.1)
Lesion
Volume (cm3), μ (σ)	9.0 (18.4)
Microadenoma, N (%)	3 (13)
Macroadenoma, N (%)	20 (87)
Secreting, N (%)
Left lesion, N (%)	10 (43.4)
Central lesion, N (%)	3 (13)
Right lesion, N (%)	14 (60.9)
Corrected T1w mean signal intensity, μ (σ)	0.03 (0.53)
Corrected T1w signal intensity heterogeneity, μ (σ)	−6.45 (14.47)
Corrected T2w mean signal intensity, μ (σ)	0.3 (1.6)
Corrected T2w signal intensity heterogeneity, μ (σ)	−3.22 (11.74)
Healthy gland
Corrected T1w mean signal intensity, μ (σ)	0.20 (0.97)
Corrected T1w signal intensity heterogeneity, μ (σ)	−2.92 (7.34)
Corrected T2w mean signal intensity, μ (σ)	0.50 (4.75)
Corrected T2w signal intensity heterogeneity, μ (σ)	0.60 (2.26)

Lesion population included three microadenomas (3/23, 13%) (<1cm), and 20 macroadenomas (20/23, 87%). Mean lesion volume was 9.0 cm³ (σ = 18.4 cm³). Ten lesions were centered to the left, 4 were centered to the right, and 9 had no side bias. Mean corrected T1W and T2W tumor SIs and SI heterogeneities are shown in Table 2.

3.3. Endocrine Status

There were six secretory pituitary adenomas: 2 corticotroph adenomas (ACTH), 2 prolactinomas, 1 somatotroph adenoma (GH), and 1 mixed prolactinoma/somatotroph adenoma. Corrected SI on T2WI was significantly higher (P=0.0196) in patients with hormone secretion (median 2.539, IQR: 0.368, 3.86) compared to those without (median: −0.378, IQR: −1.325) (P=0.004). For unit increase in corrected lesion mean SI T2, the odds of hormone secretion increased 2.76-fold. ROC curve analysis revealed 100% sensitivity and 80% specificity and AUC=0.925 for detecting hormone secretion with a corrected lesion mean SI T2 threshold of 0.06998 (Figure 5).

Figure 5.

Nominal Logistic Regression: OR 2.76 (95% CI: 1.23, 11.5). ROC Curve analysis for detecting hormone secretion: AUC = 0.925, Threshold = 0.06998, Sensitivity = 100%, Specificity = 80% (P=0.0196).

Seven patients in this study had hypopituitarism preoperatively. The curvature angle was significantly higher in subjects with preoperative hypopituitarism (median: 71.7, IQR: 36.55, 73.7) compared to those without (median: 27.25, IQR: 21.2, 41.28) (P=0.0268). For every one-degree increase in curvature angle, the odds hypopituitarism increased by 10.4% (P=0.0389) (Figure 6).

Figure 6.

Box-plot comparing pituitary stalk curvature angle in subjects with preoperative hypopituitarism (median: 71.7, IQR: 36.55, 73.7) with patients without preoperative hypopituitarism (median: 27.25, IQR: 21.2, 41.28) (P=0.0268).

4. Discussion

The present study was an exploratory examination of the utility of 7T MRI in studying the radiological correlates endocrine dysfunction in the setting of pituitary adenoma. High-spatial resolution 7T MR allowed for excellent visualization of pituitary lesional anatomy and facilitated identification of radiological markers of endocrine status. To our knowledge, ours is the first ultra-high field quantitative assessment of the entire pituitary structure, including stalk. By exploiting the high signal-to-noise ratio (SNR) and resolution at 7T to characterize sellar region features, we also developed metrics to characterize pituitary stalk structure and signal. We report three main significant findings in this study, which are discussed here. In addition to studying hormonal effects of pituitary adenomas, these methods might be applied to different endocrine disorders in the future to derive quantitative measures of stalk morphometry.

4.1. Knosp Scoring Using Ultra-high Field MRI

The present study revealed that overall ability to characterize adenomas is greater at 7T compared with lower field MRI. 7T MRI augments SNR through proportionality of signal to magnetic field strength.[16] Improved SNR can be used to enhance spatial resolution, allowing detection of tissue architecture that may not be visible at clinical field strengths.[17] Better characterization of lesions and normal glands may confer an important advantage of 7T MRI by improving definition of tumor boundaries and facilitating preservation of normal gland during surgery.

Endoscopic endonasal surgery has proven safe and efficacious in the removal of pituitary adenomas[18,19]. However, up to 10% of pituitary adenomas involve the cavernous sinus, and resultant proximity to nerves and concern for bleeding may affect degree of resectability.[4,20,21] Tumor relation to the cavernous sinus is classically quantified by Knosp scoring, which measures extent of lateral tumor extension relative to the internal carotid arteries.[22,23] An example of discordant Knosp scores found in this study is depicted in Figure 3. In this case, the left Knosp score at 3T was graded as 1 because of non-descript pituitary soft tissue crossing the medial carotid line. Increased spatial resolution and signal at 7T permitted visualization of hypointensity along the left lateral mass margin, confirming juxtaposed normal tissue, which correlated with a Knosp score of 0. Additional discordance between low field and 7T Knosp scores resulted from increased visibility of the dura at 7T (denoted by a/b subscores). Our results indicate that 7T MRI permits more accurate Knosp grading, which could confer improved accuracy in preoperative prediction of tumor invasion beyond the medial cavernous wall. Furthermore, these findings suggest that 7T imaging may improve visualization of tumors relative to other skull base compartments as well.

4.2. Higher T2 Mean SI Correlates with Secreting Tumors

In this study, secretory pituitary adenomas demonstrated significantly greater corrected T2-weighted mean SI compared with nonsecretory tumors. Prior studies have found that growth hormone-secreting adenomas that do not respond to first line somatostatin analogues tend to be hyperintense on T2-weighted imaging compared with tumors that do respond.[24,25] T2 SI of prolactinomas has also been studied and 63–80% have been classified as hyperintense on T2-weighted imaging.[5,8]

While these studies provide important evidence of an association between tumor T2 SI and endocrine function, the majority of these studies qualitatively designate tumor as hypo-/iso-/hyperintense compared with adjacent tissue. This subjective technique could suffer from operator bias. We propose a quantitative, objective method of tumor SI calculation that we believe is important in order to accurately capture biological properties of tissue. Our finding of increased corrected T2 SI in hormone-secreting tumors could reflect a property of elevated protein precursor components that may be a useful marker for stratifying patients into prognostic categories. Future studies examining tumor T2 SI to post-operative endocrine status and response to adjuvant pharmacotherapy are warranted.

4.3. Degree of Stalk Curvature Angle Correlates with Hypopituitarism

We also found that curvature angle of the pituitary stalk was significantly greater in patients who presented with hypopitutarism. There is currently no consensus regarding imaging elements that may predispose individuals to hypopituitarism, although reliable predictors could be clinically valuable. Factors such as tumor volume, secretory status, prior pituitary surgery, among others, have been shown to be poor indicators of hypopituitarism.[26]

Hypopituitarism is a multifactorial and depends on secretion status of the hypothalamus, delivery of neurohormones from the hypothalamus to the anterior lobe, and integrity of hormone-producing cells in the anterior lobe.[27] In the setting of adenomas, mass effect on the stalk by an expanding tumor can compress portal vessels, and prolonged compression may cause ischemic necrosis to the anterior lobe.[28,27] Hormone release from the anterior lobe is also dependent on transport of hypothalamic neurohormones through an intact portal system. Compression of the stalk may compromise delivery of neurohormonal signaling and result in undersecretion of anterior gland hormones. While prior literature has shown that damage to the stalk during pituitary surgery can result in postoperative hypopituitarism[29], our study suggests that stalk curvature angle may be a sensitive predictor of preoperative hypopituitarism, and may be a precise measure of mechanical compression of the pituitary stalk in patients with pituitary adenoma. Although we report significant findings, our dataset was small and future studies are needed to validate stalk curvature as a marker for preoperative hypopituitarism. Additionally, future studies to determine the association between curvature angle of the stalk after surgery and postoperative hypopituitarism status are also warranted.

4.4. Limitations

The present study was limited by sample size due to multiple contraindications that make patient recruitment criteria more stringent at 7T.[30] This limited analysis based on type of hormone secreting tumor may have reduced detectability of some findings. Additionally, since endocrine data was collected retrospectively, variable availability of lab results further limit statistical power in this study.

While 7T MRI confers higher SNR and increased resolution, it is also accompanied by more severe imaging artifacts such as B0 and B1 inhomogeneities.[31] An optimized skull base imaging protocol as well as advanced shimming techniques and dielectric pads were used to mitigate artifacts, and a neuroradiologist visually inspected images for quality assurance. Because B0 and B1 effects could not be completely eliminated, we corrected SI measurements by mathematically normalizing values to adjacent normal gray and white matter. We believe this calculation mitigated the effects of regional artifacts and enabled calculations of SI changes associated with biological processes reflecting endocrine function. Future techniques such as parallel transmit coils could be used to transmit a more uniform B1 and reduce artifact in this region.

We believe that the discrepancy in Knosp scores calculated at 7T and lower field strength reflect increased visibility of sellar anatomy, such as meningeal layers, resolved using ultra-high field MRI. However, intraoperative validation is needed to verify that the differences found in this study reflected enhanced diagnostic accuracy of Knosp scores at 7T. Future studies are warranted to determine preoperative planning benefits conferred by 7T MRI in grading lateral tumor extension and cavernous sinus involvement.

5. Conclusion

This exploratory study reports quantitative radiological features of pituitary anatomy in 23 patients with pituitary adenomas. Objective methodology for measuring precise morphometry of the pituitary stalk was developed, and may be useful in further characterizing radiological correlates of endocrine abnormality in patients with pituitary adenoma. 7T MRI was shown to yield different Knosp scores compared with lower field MRI, and also permitted enhanced characterization of lesion architecture and detection of normal gland. Lastly, we found that pituitary stalk curvature may be a potential marker of hypopituitatism, and future studies are warranted to determine the utility of this marker in the treatment and risk assessment of patients with pituitary adenomas.

Key points:

• 7T MRI yielded Knosp scores that were significantly different than those derived using lower field strength imaging.

• Secreting tumors had higher average T2-weighted SI than non-secreting tumors.

• Tumors associated with preoperative hypopituitarism had greater stalk curvature angles.

Abbreviations:

ROI

region of interest

SI

signal intensity

SNR

signal-to-noise ratio

T

Tesla