Accepted Article
Article Type: Original Article
Usefulness of dynamic MRI Enhancement measures for the diagnosis of ACTH-producing pituitary adenomas
Qinghua Guo, MD PhD 1, William F. Young, MD2, Dana Erickson, MD2, Bradley Erickson, MD PhD3* *Corresponding author
1 Division of Endocrinology, Chinese PLA General Hospital, Beijing, 100853, China 2 Division of Endocrinology, Metabolism and Nutrition, Mayo Medical School, Mayo Clinic, Rochester, MN, 55905, USA 3 Department of Radiology, Mayo Medical School, Mayo Clinic, Rochester, MN, 55905, USA
Reprints and correspondence: Name: Bradley Erickson Address: Department of Radiology, Mayo Medical School, Mayo Clinic, Rochester, MN,
55905, USA E-mail: [email protected]
Abstract
Purpose: The distinction between corticotropin (ACTH)-producing pituitary adenomas This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/cen.12475 This article is protected by copyright. All rights reserved.
Accepted Article
and occult ectopic ACTH neoplasms is a challenge and frequently complicated by the detection of nonfunctioning pituitary adenomas on dynamic magnetic resonance imaging (DMRI). Herein, we explored quantitative differences in DMRI measures from ACTH-producing pituitary adenomas.
Materials and Methods: Fifty-five patients with pathology confirmed ACTH-producing and 44 with nonfunctioning pituitary adenomas were analyzed in this retrospective pilot study. The intensities of adenomas and of adjacent pituitary tissue were measured by drawing a region of interest. Time–intensity curves were then constructed and quantitative analysis included: enhancement time, enhancement peak, and pre-peak slope (PPS). Multivariable logistic analysis and receiver operating characteristic curves (ROC) were used to evaluate the parameters. Results: Quantitative analysis showed that tumor PPS in ACTH-producing adenomas was markedly lower than that in nonfunctioning adenomas (P=0.0042) and that, the PPS of the adjacent pituitary gland were not different (P=0.2441). The PPS was > 3.0 in 59.1% of nonfunctioning adenomas and ≤ 3.0 in 69.1% of ACTH-producing adenomas (P=0.0049). Logistic analysis revealed lower tumor PPS levels (P=0.0123, OR 1.24, 95% CI: 1.05-1.52) were associated with ACTH-producing adenomas. The optimal PPS cutoff determined by ROC analysis was 2.89, with a sensitivity of 69% and specificity of 70%. No significant difference was found between the two groups in enhancement time or enhancement peak. Conclusion: Enhancement parameters in DMRI can help distinguishing ACTHproducing from nonfunctioning pituitary adenomas, which could be useful in the differential diagnosis between ACTH-producing pituitary adenomas and ectopic ACTH syndrome complicated with nonfunctioning pituitary adenoma.
This article is protected by copyright. All rights reserved.
Accepted Article
Introduction Although Cushing syndrome (CS) was first described more than 100 years ago 1,
there remain many diagnostic and therapeutic challenges. Distinguishing between corticotropin (ACTH)-producing pituitary adenoma (Cushing disease) and occult ectopic ACTH syndrome (EAS) can be problematic 2, 3. Although severe and rapidly progressing ACTH-dependent CS can be relatively easily attributed to EAS, those patients with EAS with mild presentations pose a great challenge. The most frequent etiology of EAS was slow-growing bronchial carcinoid tumors (25% of EAS) and various other carcinoid tumors (53% of EAS) 4. Importantly, the differential diagnosis between indolent EAS and
ACTH-secreting pituitary adenoma is complicated by the fact that nonfunctioning pituitary adenoma5-7 and pituitary incidentalomas are found in 10-20% of in the general population 8, 9. Several methods, including plasma ACTH, high-dose dexamethasone suppression
test, corticotropin releasing hormone stimulation test, and inferior petrosal sinus sampling (IPSS), have been used to distinguish between pituitary-dependent CS and EAS
2, 10
. However, these tests have some limitations due to suboptimal sensitivity and
specificity11, 12. While IPSS has excellent sensitivity and specificity
13
, it must be noted
that it is an invasive test with potential risks and serious reported complications14, 15. MRI with gadolinium enhancement has been widely used to detect pituitary tumors in
the setting of ACTH-dependent CS
3, 16
. The finding of an apparent pituitary adenoma
on MRI in a patient with CS is complicated by the knowledge that pituitary incidentalomas have been demonstrated in 10-38% of unselected adult patients with MRI
17, 18
and the detection prevalence is increasing with the new generation MR
This article is protected by copyright. All rights reserved.
Accepted Article
scanners
3, 8, 16
. Interpretation of all of the previously described diagnostic tests
becomes more challenging when ectopic ACTH syndrome may be complicated with a nonfunctioning pituitary adenoma. Recently, it was reported that quantitative or semiquantitative parameters obtained
with dynamic MRI (DMRI) are correlated with the degree of angiogenesis in various malignancies and had been successfully used in the detection of abnormal subendocardial perfusion
19
, the prognostic analysis of rectal cancer20, as well as
evaluation of cerebral intraventricular tumors21. Pituitary adenoma subtypes have different vascular characters22 and ACTH-secreting tumor was found to be less
vascular23. In the current study, we proposed a hypothesis that DMRI could prove to be useful in
the differentiation of different subtypes of adenomas with different vascular characteristics. Herein, we reviewed our experience with dynamically acquired MRI to determine if MRI enhancement measures could be used to distinguish between ACTHsecreting and nonfunctioning pituitary tumors.
Materials and Methods Patients After Institutional Review Board approval, we searched our radiology database to
retrieve all the patients with visible lesions of pituitary adenomas in DMRI between 1997 and 2012 and undertook a retrospective pilot study. Fifty-five patients with ACTHproducing pituitary microadenomas (≤10 mm) and 44 nonfunctioning pituitary
This article is protected by copyright. All rights reserved.
Accepted Article
microadenomas were identified. The patients with ACTH-producing adenomas had clinical and biochemical Cushing syndrome and were treated with transsphenoidal surgery and tumor subtype was confirmed with immunohistochemistry. A pituitary adenoma was considered as nonfunctioning when it met the modified criteria provided by Karavitaki
24
: (i) imaging features suggestive of a pituitary adenoma; (ii) no clinical
and/or biochemical evidence of hormonal hypersecretion; (iii) lack of function on repeat monitoring over at least 2 years (evaluation included repeat imaging, clinical assessment, and biochemical evaluation). All patients were evaluated by an endocrinologist in follow-up examinations and all had DMRI with visible lesions on the images. Patients treated previously with external radiation, and those with negative MRI findings were excluded. MRI Images were acquired with 1.5T or 3T MRI with thin (≤ 3 mm) coronal sections
through the sella before and after gadolinium administration. The protocol included dynamic T1-weighted fast spin echo sequence through the sella during the injection of 10 cc gadolinium (0.05 mmol/kg) followed by 40 cc of normal saline at a standard rate of 3 cc/s. Four slice locations were acquired every 20-30 seconds. This sequence was TR/TE 400/min full echo, with an ETL of 3, a slice thickness of 3 mm with 0.3 mm interslice gap. Tumor size (mm) was measure by maximum diameter. Quantitative Analysis and Calculations of dynamic images 1) Time–intensity curve The intensities of adenoma and adjacent pituitary gland in subjects at each time point
in the sequence were measured by drawing a region of interest (ROI), which included
This article is protected by copyright. All rights reserved.
Accepted Article
as much of the adenoma (A) as possible and a matching or larger normal anterior lobe pituitary gland (P) on the dynamic sequence image that best demonstrated the tumor, as shown in Figure 1. The ROIs were manually traced by trained reader with review by the expert neuroradiologist who designed the program. They were blinded to clinical information until after all tracings were completed. The mean areas of ROI for tumor were 5.5 ± 1.9 mm2 and 11.3 ± 3.4 mm2 for pituitary. Those ROIs were then propagated
to the images at the same location but other time points in the DMRI acquisition and the respective mean intensities calculated using custom software. Then, time–intensity curves were constructed from signal intensity values obtained from the ROIs. 2) Enhancement parameters The enhancement parameters were obtained and calculated from signal intensity
values of the defined ROIs. Sequential quantification of dynamic images included: enhancement ratio (ER), enhancement time (ET), enhancement peak (EP), and prepeak slope (PPS), which were calculated and expressed as follows: ① ER=P (t)/P (0) × 100 or A (t)/P (0) × 100, where: P (t), intensity of pituitary at time
t; P (0), intensity of pituitary at time 0; A (t), intensity of adenoma at time t. ② Enhancement time (ET): defined as the relative time that adenomas reached their
peak enhancement relative to normal gland: early, simultaneous, or late. ③ Enhancement peak (EP): the highest enhancement intensity of pituitary or
adenoma ④ Pre-peak slope (PPS): referring to the measure of the tissue perfusion rate or
speed reaching peak intensity in the enhancement curve. It was calculated as PPS=E1-
This article is protected by copyright. All rights reserved.
Accepted Article
E2/T1-T2, where: E1, 75% peak intensity of pituitary or adenoma; E2, intensity at time point prior to reaching 75% peak intensity; T1, time of E1; T2, time of E2.
Statistical Analyses Continuous clinical and biologic variables were described with mean ± standard
deviation or mean ± standard error of the mean. Categorical variables were expressed as number and percentage. The PPS was stratified into 3 groups with the criteria: 03.0. Comparison of quantitative data: age, tumor size, peak intensity, and PPS etc. between groups were analyzed using Student’s t-test, or Wilcoxon rank-sum
when necessary, while
comparison of sex, adenoma location in pituitary gland, enhancement time, and PPS category variables were analyzed using Chi-square test or Fisher`s exact test, which were all carried out with the JMP 9 package. Multivariable Logistic regression analysis was used to analyze the contributing factors to the adenoma categories. Receiver operating characteristic (ROC) curves were used to evaluate the diagnostic potential of tumor PPS used for semiquantitative assessment. Area under the curve (AUC) was determined and cutoff value chosen that maximized the Youden-criterion for the indicator. Sensitivities, specificities, and diagnostic accuracies were determined with regard to the cutoff. Results were considered statistically significant if the two-tailed Pvalue was
Article Type: Original Article
Usefulness of dynamic MRI Enhancement measures for the diagnosis of ACTH-producing pituitary adenomas
Qinghua Guo, MD PhD 1, William F. Young, MD2, Dana Erickson, MD2, Bradley Erickson, MD PhD3* *Corresponding author
1 Division of Endocrinology, Chinese PLA General Hospital, Beijing, 100853, China 2 Division of Endocrinology, Metabolism and Nutrition, Mayo Medical School, Mayo Clinic, Rochester, MN, 55905, USA 3 Department of Radiology, Mayo Medical School, Mayo Clinic, Rochester, MN, 55905, USA
Reprints and correspondence: Name: Bradley Erickson Address: Department of Radiology, Mayo Medical School, Mayo Clinic, Rochester, MN,
55905, USA E-mail: [email protected]
Abstract
Purpose: The distinction between corticotropin (ACTH)-producing pituitary adenomas This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/cen.12475 This article is protected by copyright. All rights reserved.
Accepted Article
and occult ectopic ACTH neoplasms is a challenge and frequently complicated by the detection of nonfunctioning pituitary adenomas on dynamic magnetic resonance imaging (DMRI). Herein, we explored quantitative differences in DMRI measures from ACTH-producing pituitary adenomas.
Materials and Methods: Fifty-five patients with pathology confirmed ACTH-producing and 44 with nonfunctioning pituitary adenomas were analyzed in this retrospective pilot study. The intensities of adenomas and of adjacent pituitary tissue were measured by drawing a region of interest. Time–intensity curves were then constructed and quantitative analysis included: enhancement time, enhancement peak, and pre-peak slope (PPS). Multivariable logistic analysis and receiver operating characteristic curves (ROC) were used to evaluate the parameters. Results: Quantitative analysis showed that tumor PPS in ACTH-producing adenomas was markedly lower than that in nonfunctioning adenomas (P=0.0042) and that, the PPS of the adjacent pituitary gland were not different (P=0.2441). The PPS was > 3.0 in 59.1% of nonfunctioning adenomas and ≤ 3.0 in 69.1% of ACTH-producing adenomas (P=0.0049). Logistic analysis revealed lower tumor PPS levels (P=0.0123, OR 1.24, 95% CI: 1.05-1.52) were associated with ACTH-producing adenomas. The optimal PPS cutoff determined by ROC analysis was 2.89, with a sensitivity of 69% and specificity of 70%. No significant difference was found between the two groups in enhancement time or enhancement peak. Conclusion: Enhancement parameters in DMRI can help distinguishing ACTHproducing from nonfunctioning pituitary adenomas, which could be useful in the differential diagnosis between ACTH-producing pituitary adenomas and ectopic ACTH syndrome complicated with nonfunctioning pituitary adenoma.
This article is protected by copyright. All rights reserved.
Accepted Article
Introduction Although Cushing syndrome (CS) was first described more than 100 years ago 1,
there remain many diagnostic and therapeutic challenges. Distinguishing between corticotropin (ACTH)-producing pituitary adenoma (Cushing disease) and occult ectopic ACTH syndrome (EAS) can be problematic 2, 3. Although severe and rapidly progressing ACTH-dependent CS can be relatively easily attributed to EAS, those patients with EAS with mild presentations pose a great challenge. The most frequent etiology of EAS was slow-growing bronchial carcinoid tumors (25% of EAS) and various other carcinoid tumors (53% of EAS) 4. Importantly, the differential diagnosis between indolent EAS and
ACTH-secreting pituitary adenoma is complicated by the fact that nonfunctioning pituitary adenoma5-7 and pituitary incidentalomas are found in 10-20% of in the general population 8, 9. Several methods, including plasma ACTH, high-dose dexamethasone suppression
test, corticotropin releasing hormone stimulation test, and inferior petrosal sinus sampling (IPSS), have been used to distinguish between pituitary-dependent CS and EAS
2, 10
. However, these tests have some limitations due to suboptimal sensitivity and
specificity11, 12. While IPSS has excellent sensitivity and specificity
13
, it must be noted
that it is an invasive test with potential risks and serious reported complications14, 15. MRI with gadolinium enhancement has been widely used to detect pituitary tumors in
the setting of ACTH-dependent CS
3, 16
. The finding of an apparent pituitary adenoma
on MRI in a patient with CS is complicated by the knowledge that pituitary incidentalomas have been demonstrated in 10-38% of unselected adult patients with MRI
17, 18
and the detection prevalence is increasing with the new generation MR
This article is protected by copyright. All rights reserved.
Accepted Article
scanners
3, 8, 16
. Interpretation of all of the previously described diagnostic tests
becomes more challenging when ectopic ACTH syndrome may be complicated with a nonfunctioning pituitary adenoma. Recently, it was reported that quantitative or semiquantitative parameters obtained
with dynamic MRI (DMRI) are correlated with the degree of angiogenesis in various malignancies and had been successfully used in the detection of abnormal subendocardial perfusion
19
, the prognostic analysis of rectal cancer20, as well as
evaluation of cerebral intraventricular tumors21. Pituitary adenoma subtypes have different vascular characters22 and ACTH-secreting tumor was found to be less
vascular23. In the current study, we proposed a hypothesis that DMRI could prove to be useful in
the differentiation of different subtypes of adenomas with different vascular characteristics. Herein, we reviewed our experience with dynamically acquired MRI to determine if MRI enhancement measures could be used to distinguish between ACTHsecreting and nonfunctioning pituitary tumors.
Materials and Methods Patients After Institutional Review Board approval, we searched our radiology database to
retrieve all the patients with visible lesions of pituitary adenomas in DMRI between 1997 and 2012 and undertook a retrospective pilot study. Fifty-five patients with ACTHproducing pituitary microadenomas (≤10 mm) and 44 nonfunctioning pituitary
This article is protected by copyright. All rights reserved.
Accepted Article
microadenomas were identified. The patients with ACTH-producing adenomas had clinical and biochemical Cushing syndrome and were treated with transsphenoidal surgery and tumor subtype was confirmed with immunohistochemistry. A pituitary adenoma was considered as nonfunctioning when it met the modified criteria provided by Karavitaki
24
: (i) imaging features suggestive of a pituitary adenoma; (ii) no clinical
and/or biochemical evidence of hormonal hypersecretion; (iii) lack of function on repeat monitoring over at least 2 years (evaluation included repeat imaging, clinical assessment, and biochemical evaluation). All patients were evaluated by an endocrinologist in follow-up examinations and all had DMRI with visible lesions on the images. Patients treated previously with external radiation, and those with negative MRI findings were excluded. MRI Images were acquired with 1.5T or 3T MRI with thin (≤ 3 mm) coronal sections
through the sella before and after gadolinium administration. The protocol included dynamic T1-weighted fast spin echo sequence through the sella during the injection of 10 cc gadolinium (0.05 mmol/kg) followed by 40 cc of normal saline at a standard rate of 3 cc/s. Four slice locations were acquired every 20-30 seconds. This sequence was TR/TE 400/min full echo, with an ETL of 3, a slice thickness of 3 mm with 0.3 mm interslice gap. Tumor size (mm) was measure by maximum diameter. Quantitative Analysis and Calculations of dynamic images 1) Time–intensity curve The intensities of adenoma and adjacent pituitary gland in subjects at each time point
in the sequence were measured by drawing a region of interest (ROI), which included
This article is protected by copyright. All rights reserved.
Accepted Article
as much of the adenoma (A) as possible and a matching or larger normal anterior lobe pituitary gland (P) on the dynamic sequence image that best demonstrated the tumor, as shown in Figure 1. The ROIs were manually traced by trained reader with review by the expert neuroradiologist who designed the program. They were blinded to clinical information until after all tracings were completed. The mean areas of ROI for tumor were 5.5 ± 1.9 mm2 and 11.3 ± 3.4 mm2 for pituitary. Those ROIs were then propagated
to the images at the same location but other time points in the DMRI acquisition and the respective mean intensities calculated using custom software. Then, time–intensity curves were constructed from signal intensity values obtained from the ROIs. 2) Enhancement parameters The enhancement parameters were obtained and calculated from signal intensity
values of the defined ROIs. Sequential quantification of dynamic images included: enhancement ratio (ER), enhancement time (ET), enhancement peak (EP), and prepeak slope (PPS), which were calculated and expressed as follows: ① ER=P (t)/P (0) × 100 or A (t)/P (0) × 100, where: P (t), intensity of pituitary at time
t; P (0), intensity of pituitary at time 0; A (t), intensity of adenoma at time t. ② Enhancement time (ET): defined as the relative time that adenomas reached their
peak enhancement relative to normal gland: early, simultaneous, or late. ③ Enhancement peak (EP): the highest enhancement intensity of pituitary or
adenoma ④ Pre-peak slope (PPS): referring to the measure of the tissue perfusion rate or
speed reaching peak intensity in the enhancement curve. It was calculated as PPS=E1-
This article is protected by copyright. All rights reserved.
Accepted Article
E2/T1-T2, where: E1, 75% peak intensity of pituitary or adenoma; E2, intensity at time point prior to reaching 75% peak intensity; T1, time of E1; T2, time of E2.
Statistical Analyses Continuous clinical and biologic variables were described with mean ± standard
deviation or mean ± standard error of the mean. Categorical variables were expressed as number and percentage. The PPS was stratified into 3 groups with the criteria: 03.0. Comparison of quantitative data: age, tumor size, peak intensity, and PPS etc. between groups were analyzed using Student’s t-test, or Wilcoxon rank-sum
when necessary, while
comparison of sex, adenoma location in pituitary gland, enhancement time, and PPS category variables were analyzed using Chi-square test or Fisher`s exact test, which were all carried out with the JMP 9 package. Multivariable Logistic regression analysis was used to analyze the contributing factors to the adenoma categories. Receiver operating characteristic (ROC) curves were used to evaluate the diagnostic potential of tumor PPS used for semiquantitative assessment. Area under the curve (AUC) was determined and cutoff value chosen that maximized the Youden-criterion for the indicator. Sensitivities, specificities, and diagnostic accuracies were determined with regard to the cutoff. Results were considered statistically significant if the two-tailed Pvalue was
