M. Godlewska*a (Ms), A. Grochowskaa (Dr), A. BogusÅ‚awskaa (Ms), A. Hubalewska-Dydejczyka (Prof), A. Gilis-Januszewskaa (Prof)

a Jagiellonian University Medical College, Cracow, POLAND

* magda.godlewska92@gmail.com


Volumetric assessment is not a standard procedure in the analysis of pituitary Magnetic Resonance Imaging (MRI), even though it is reported to be reliable and might predict the surgical outcomes.


Our aim was to evaluate the outcomes of two quantitative methods of tumor volume measurements in patients with acromegaly.

Material and methods

72 consecutive patients diagnosed with acromegaly between 01.01.2012 and 31.12.2021 at the Department of Endocrinology, Jagiellonian University Medical College were evaluated. Inclusion criteria were: biochemically confirmed acromegaly before the surgery or pharmacological treatment, sellar tumor confirmed in the pituitary MRI, the MRI had to be performed before the surgery and implementation of any medical treatment. 58 patients were found eligible for the study. Pituitary MR images were analysed by a specialist of radiology with experience in the interpretation of the sellar MRI. Retrospectively, the tumor volume was measured by manual segmentation (Method 1) and by manual delineation of volume of interest (VOI)(Method 2). Other radiological parameters noted included: the largest tumor diameter, parasellar expansion assessed using modified Knosp classification, evaluation for the compression of the optic chiasm and signs of the intratumoral bleeding.


The median TV was 1.02 cm3 (IQR 2.08) for Method 1 and 1.23 cm3 (IQR 2.21) for Method 2. There was a strong correlation between measurements in two Methods (R=0.949; p<0.01) as well as between each method and the maximal tumor diameter (R=0.884, p<0.001 for Method 1 and R=0.893; p<0.01 for Method 2). Univariate logistic regression showed, that in both Methods larger tumor size was associated with higher risk of optic chiasm compression and invasion of the cavernous sinuses. Neither method was associated with tumor infarction. Surprisingly, tumor volume estimation using both methods did not predict the outcomes of the surgery in the univariate logistic regression (p=0.117)


Both methods show a strong correlation between each other and with median tumor diameter- therefore their use may be reliable. Surprisingly, tumor volume did not predict the neurosurgery outcomes in our cohort of patients.

The author has declared no conflict of interest.