Imaging Errors in Distinguishing Pituitary Adenomas From Other Sellar Lesions : Journal of Neuro-Ophthalmology

Secondary Logo

Journal Logo

Original Contribution

Imaging Errors in Distinguishing Pituitary Adenomas From Other Sellar Lesions

Altshuler, David B. MD; Andrews, Chris A. PhD; Parmar, Hemant A. MD; Sullivan, Stephen E. MD; Trobe, Jonathan D. MD

Editor(s): Fraser, Clare MD; Mollan, Susan MD

Author Information
Journal of Neuro-Ophthalmology 41(4):p 512-518, December 2021. | DOI: 10.1097/WNO.0000000000001164
  • Free

Abstract

Although pituitary adenomas account for more than 90% of sellar region masses, nonadenomatous neoplastic, developmental, inflammatory, and vascular lesions also occur at this site (1–27). If nonadenomatous lesions are misinterpreted as pituitary adenomas on imaging, inappropriate surgical decision-making may result. For example, in sellar metastasis, a systemic evaluation would be warranted; in craniopharyngioma, the surgical approach would differ from that for adenoma; in inflammatory lesions, surgery would not be undertaken at all.

The fact that errors occur in distinguishing adenomas from nonadenomatous lesions is well known to radiologists and clinicians. But the types of errors have not been as well documented. Accordingly, we designed a study in which 2 skull-base neurosurgeons and 2 neuroradiologists of differing experience levels were asked to interpret deidentified MRI studies of sellar adenomas and nonadenomas. The 4 readers were asked to make diagnoses, explain which imaging features they used to reach a differential diagnosis, and their confidence in doing so. We analyzed the errors and the types of lesions on which those errors were made.

METHODS

We included 18 nonadenomatous sellar region masses from a list of cases presenting to Michigan Medicine between 2015 and 2019 and collected by 2 authors (H.A.P. and J.D.T.) because at least one primary radiologist had mistakenly diagnosed a pituitary adenoma. We also selected 21 pituitary adenomas from the Michigan Medicine radiology files in which at least 1 primary radiologist had made the correct imaging diagnosis. Surgical specimens had confirmed the pathology in all cases (Table 1).

TABLE 1. - Pathologic diagnoses of sellar lesions
Pathologic Diagnosis Total Cases
Pituitary adenoma 21
Nonadenomatous lesion 18
 Craniopharyngioma 3
 Germinoma 1
 Meningioma 2
 Metastasis 3
 Langerhans cell histiocytosis 1
 Rathke cleft cyst 2
 Plasmacytoma 1
 Sinonasal undifferentiated carcinoma 2
 Glioma 1
 Chordoma 1
 Aneurysm 1

Four readers were instructed to interpret the imaging studies. There were 2 skull-base subspecialty neurosurgeons and 2 neuroradiologists. For each pair, 1 reader was selected as having had more years of experience in the field, and 1 reader was selected as having had less experience. The less experienced neurosurgeon had had 14 years of postgraduate experience; the senior neurosurgeon had had 39 years. The less experienced neuroradiologist had had 3 years of postgraduate experience in imaging interpretation; the more experienced neuroradiologist had had 15 years. The readers were masked as to patient identity and diagnosis. They were given compact discs containing MRIs with precontrast and postcontrast T1-weighted axial and coronal sequences and T2-weighted axial and sagittal/coronal reconstructed MRI sequences. Readers were told that they were viewing adenoma and nonadenoma sellar region cases but not how many of each. They were notified in advance that there were an unequal number of adenoma and nonadenoma cases to prevent them from arriving at a diagnosis by tallying their previous diagnoses.

The readers' responses were recorded on a standard questionnaire that called for them to decide if the lesions were adenomas or nonadenomas and to list the imaging criteria used to reach that conclusion. If their imaging diagnosis was nonadenoma, they were asked to list a differential diagnosis in order of likelihood. Finally, they were asked to rate their confidence in diagnosis on a 5-point Likert scale: “not confident,” “slightly confident,” “moderately confident,” “very confident,” or “extremely confident.”

In analyzing the readers' responses, we noted whether they were correct or incorrect in differentiating adenoma from nonadenoma. If they chose nonadenoma, we noted the rank order of nonadenoma diagnoses, which imaging characteristics were listed as contributory to their diagnoses, and the confidence with which readers made their diagnoses. Where there were diagnostic errors, we determined whether there were imaging features that should have allowed a correct diagnosis (“avoidable errors”) and where no distinguishing imaging features were present (“unavoidable errors”).

Data were collected in Qualtrics (Provo, UT) and summarized in R (Vienna, Austria).

The study was approved by our institutional review board.

RESULTS

The 4 readers contributed 156 “readings” of the 39 cases. Among the 18 nonadenoma cases, 11 (61%) were incorrectly diagnosed by at least 1 reader. Among the 21 adenoma cases, 8 (38%) were incorrectly diagnosed by at least 1 reader.

Interrater agreement was relatively high, with Fleiss's kappa among all 4 readers equal to 0.54. The overall diagnostic accuracy between neurosurgeons and neuroradiologists did not differ significantly. The experience level was also an insignificant contributor to diagnostic accuracy.

Readers reported that they were “very confident” or “extremely confident” in their diagnoses in 58% of readings. Only 7% of readings were made with “slight confidence” or “no confidence.” The distribution of diagnostic confidence did not differ significantly between adenomas and nonadenomas. The proportion of correct diagnoses increased with the level of confidence, but 13% of incorrect diagnoses were made with “high confidence.”

Among the 7 nonadenoma cases correctly diagnosed by all 4 readers, identification of the pituitary as separate from the lesion was used as the criterion (Fig. 1).

F1
FIG. 1.:
Sellar craniopharyngioma and meningioma diagnosed correctly. A. Sagittal postcontrast enhanced T1 weighted image shows a cystic suprasellar lesion (craniopharyngioma) with a small enhancing solid component (arrowhead). In making the correct diagnosis, all 4 readers noted that the sella was not enlarged and that the mass was clearly separate from the pituitary gland (arrow). B. Sagittal post-contrast T1-weighted image in another patient shows an enhancing mass (arrowheads) in the suprasellar region (meningioma). In making the correct diagnosis, all 4 readers noted that the pituitary gland (arrow) was separate from the mass and had a different enhancement pattern. They also noted that the sella was not enlarged.

Among the 11 nonadenoma cases misdiagnosed as adenomas by at least 1 reader, avoidable errors occurred in 5 cases (Table 2). In 4 of them, the error was a failure to appreciate that the lesion was separate from the pituitary gland. A separate gland was visible on all pulse sequences in 3 of those cases, but in 1 case—the plasmacytoma—scrutiny of the precontrast T1 sequence was necessary to make that observation (Fig. 2). A further avoidable error occurred in the aneurysm case: 1 reader failed to appreciate the lamellated appearance with a markedly hypointense signal on the T2 sequence (Fig. 3) that would be characteristic of an aneurysm. In the remaining 6 misdiagnosed nonadenoma cases, the errors were considered unavoidable in 5 cases. In 4 of those unavoidable errors, the distinguishing features had been destroyed by the lesion, including a chordoma, sphenoid sinus carcinoma (Fig. 4), metastasis (2 cases), and an undifferentiated sinonasal carcinoma. In 1 case—a Langerhans cell histiocytosis—the unavoidable error occurred because the lesion had features that made it indistinguishable from a cystic pituitary adenoma.

TABLE 2. - The eleven nonadenoma cases misdiagnosed as adenoma by at least 1 reader
Pathologic Diagnosis No. of Readers Who Diagnosed Adenoma Avoidable Errors Unavoidable Errors
Rathke cleft cyst 1 Lesion is separate from the pituitary gland on several pulse sequences
Plasmacytoma (Fig. 2) 2 Lesion is separate from the pituitary gland on precontrast T1 sequence
Aneurysm ( Fig. 3 ) 1 Lamellated hypointense T2 signal is not characteristic of adenoma
Craniopharyngioma 1 Lesion is separate from the pituitary gland on several pulse sequences
Chordoma 1 Lesion is separate from the pituitary gland on several pulse sequences
Chordoma 3 Large lesion destroys landmarks that allow identification of the pituitary gland as separate
Langerhans cell histiocytosis 1 Features mimic cystic adenoma
Sphenoid sinus carcinoma (Fig. 4) 4 Large lesion destroys landmarks that allow identification of the pituitary gland as separate
Metastasis 4 Large lesion destroys landmarks that allow identification of the pituitary gland as separate
Metastasis 4 Large lesion destroys landmarks that allow identification of the pituitary gland as separate
Sinonasal undifferentiated carcinoma 1 Large lesion destroys landmarks that allow identification of the pituitary gland as separate

F2
FIG. 2.:
Plasmacytoma misdiagnosed as pituitary adenoma. Sagittal precontrast (A) and postcontrast (B) T1-weighted images show a large, slightly heterogeneously enhancing mass (arrows, B; arrowheads, A) in the sellar–suprasellar region that was misdiagnosed by 2 of 4 readers as a pituitary adenoma. However, the precontrast T1-weighted image (A) shows that the pituitary gland (arrow) is superior to the mass (arrowheads), which should have been recognized as allowing a diagnosis of a nonadenomatous lesion.
F3
FIG. 3.:
Intrasellar aneurysm misdiagnosed as pituitary adenoma. Sagittal postcontrast T1-weighted image (A) shows a hypointense, nonenhancing, lamellated signal (arrows). Coronal T2-weighted image (B) shows an extremely hypointense sellar mass. One reader diagnosed a pituitary adenoma, an error that had also been made on preoperative interpretation. Fortunately, in the operating room, the surgeon recognized this signal abnormality as concerning for an aneurysm and halted the surgery. Catheter angiography later identified a thrombosed aneurysm from the internal carotid artery.
F4
FIG. 4.:
Sphenoid sinus carcinoma misdiagnosed as pituitary adenoma. Sagittal precontrast (A) and postcontrast (B) T1-weighted images show a slightly heterogeneous large mass in the sellar–suprasellar region that has destroyed normal landmarks (arrows). All 4 readers incorrectly identified this lesion as a pituitary adenoma, perhaps because the pituitary gland was not seen as separate from the mass. When a lesion obliterates telltale features, the distinction between adenoma and nonadenoma becomes difficult.

Among the 21 adenoma cases, the diagnosis was made correctly by all 4 readers in 13 (62%) cases. Readers used 2 imaging characteristics to diagnose adenoma: (1) the mass was inseparable from the pituitary gland and (2) the sella was enlarged (Fig. 5). Less commonly used criteria were homogeneous enhancement of the mass (78.6%) and cavernous sinus invasion (28.6%).

F5
FIG. 5.:
Adenoma diagnosed correctly. Sagittal (A) and coronal (B) postcontrast T1-weighted images show the classic imaging appearance of a pituitary macroadenoma (arrowhead), a homogeneously-enhancing solid mass in the sella–suprasellar region inseparable from the pituitary gland; the sella turcica is enlarged (arrows). The coronal image (B) shows the typical “snowman” sign of pituitary macroadenoma because of constriction of the mass at the level of the diaphragma sellae. All 4 readers correctly used these features to identify pituitary adenoma.

Among the 8 (38%) adenoma cases in which at least 1 reader made an incorrect diagnosis (Table 3), avoidable errors occurred in 7. In 2 of them, an adenoma was called a craniopharyngioma, but readers could have made the correct diagnosis by noting an enlarged sella, and, in 1 case, by realizing that hyperintensity on the precontrast T1 sequence could be from blood or protein in a pituitary adenoma (Fig. 6). Two cases were misdiagnosed as Rathke cleft cyst, but the rim was too thick for that condition. Another adenoma case was labeled a sinus carcinoma, but sellar enlargement and a sellar origin of the lesion marked it as more likely to be an adenoma. Hemangioma was given as a diagnosis by 1 reader, who failed to note that there was no separate pituitary gland and that the sella was enlarged. Pituitary hyperplasia was chosen by 1 reader, yet the sella was clearly enlarged and the lesion was too large for hyperplasia. An unavoidable error occurred in just 1 case, in which there was a large lesion with heterogeneous signal and a mixture of solid and cystic elements, justifiably warranting a diagnosis of craniopharyngioma (Fig. 7).

TABLE 3. - Eight adenoma cases misdiagnosed as nonadenoma by at least 1 reader
Misdiagnosis No. of Readers Who Made This Misdiagnosis Avoidable Errors Unavoidable Errors
Craniopharyngioma 1 Sellar enlargement, which is rare for craniopharyngioma and more characteristic of adenoma
Craniopharyngioma (Fig. 6) 2 Sellar enlargement, which is rare for craniopharyngioma; hyperintense precontrast T1 signal of blood should be recognized as occurring in adenoma; “snowman” configuration, a feature more characteristic of adenoma
Craniopharyngioma (Fig. 7) 2 Combination of solid and cystic features mimics craniopharyngioma
Rathke cleft cyst 1 Thick enhancing rim, which is rare for Rathke cleft cyst and more characteristic of adenoma
Rathke cleft cyst 1 Thick enhancing rim, which is rare for Rathke cleft cyst and more characteristic of adenoma
Sinonasal undifferentiated carcinoma 1 Arises from sella and causes sellar enlargement, features more characteristic of adenoma
Hemangioblastoma 1 Sellar enlargement and no visible pituitary gland, features more characteristic of adenoma
Pituitary hyperplasia 1 Lesion is large and causes sellar enlargement, features more characteristic of adenoma

F6
FIG. 6.:
Adenoma misdiagnosed as craniopharyngioma. Sagittal (A) and coronal (B) precontrast T1-weighted images show a large mass (arrows) in the sellar–suprasellar region that is hyperintense because of internal protein or hemorrhage. Coronal image shows the “snowman” sign typical of pituitary macroadenoma. Two readers understandably labeled this a craniopharyngioma because they believed the lesion was separate from the residual pituitary gland, which appears towards the left (arrowhead, B).
F7
FIG. 7.:
Adenoma misdiagnosed as craniopharyngioma. Sagittal precontrast (A) and postcontrast (B) T1-weighted images show a solid, enhancing mass (arrowhead, A) with precontrast T1 hyperintense cystic components (arrows, A) in the sellar–suprasellar region. Two readers misdiagnosed this lesion as a craniopharyngioma. The imaging features in this case so resembled those of craniopharyngioma that this misdiagnosis was considered unavoidable.

DISCUSSION

Our study has demonstrated that errors in differentiating adenomas from nonadenomas by imaging alone can occur even among experienced neurosurgeons and neuroradiologists. Readers from each specialty had equal error rates and experience seemed to make no difference.

We do not claim that the error frequency demonstrated in this study has “real world” validity for several reasons. First, the nonadenoma cases represent a much higher proportion of sellar lesions than is encountered in clinical practice. Second, they were selected because at least 1 radiologist had misdiagnosed them as adenomas, so they would have inordinately confounding imaging features. Third, our sample of readers was small. Fourth, readers interpreted images without any clinical information.

More valuable is the information about the nature of the errors. In the misdiagnosis of 11 (61%) of the 18 nonadenomatous lesions as adenomas, the errors were avoidable in 5 of those 11 cases. Readers failed to identify a pituitary gland separate from the lesion, a feature that sometimes required scrutiny of all available MRI sequences. In 1 case, a potentially serious avoidable error occurred when readers failed to note the lamellated, low-intensity T2 signal of an intrasellar aneurysm. In the 2 cases in which sellar metastases were misinterpreted as adenomas, the surgeon had originally been placed in the awkward position of having to disclose the true pathology days after having comforted the patient that an adenoma had been successfully removed. Interpretative errors were unavoidable in 6 of 11 misdiagnosed nonadenomatous cases when the lesion had grown to a size that the pituitary gland had been obliterated or where a nonadenomatous lesion (Langerhans cell histiocytosis) had cystic features known to be a feature of some adenomas.

In correctly identifying adenomas, readers were guided by noting that the masses were indistinguishable from the pituitary gland. This feature was most easily identifiable on the T1 postcontrast sequences. An enlarged sella and the typical “snowman” configuration were also helpful findings in diagnosing adenoma. Readers usually failed to recognize adenomas when they overlooked an enlarged sella or failed to realize that a thick rind in a cystic lesion would be more typical of adenoma than Rathke cleft cyst.

We draw the following conclusions from this study:

  1. MRI interpretative accuracy in identifying a nonadenomatous lesion could be improved by looking beyond the postcontrast sequences to clarify that the lesion appears to be separate from the pituitary gland
  2. Sellar enlargement, homogeneous enhancement, and contiguous invasion of the cavernous sinus are contributory features in identifying adenoma but are less helpful in excluding nonadenomatous lesions than noting that the lesion arises within the pituitary gland
  3. When lesions are large, they may markedly displace or even obliterate the pituitary gland, making differentiation of adenoma from nonadenoma difficult unless the reader looks carefully for a displaced gland remnant or notes that the origin of the lesion is extrasellar
  4. Cystic adenomas can look like craniopharyngiomas or Rathke cleft cysts, making the distinction impossible

The principal contribution of this clinical experiment is to document where interpretative mistakes may occur in imaging diagnosis of sellar lesions and how they might be avoided. It found that readers of MRIs could improve their accuracy by carefully examining all the available imaging data to determine the relationship of the lesion to the pituitary gland, acknowledging that there are some cases in which that distinguishing feature may be lost. Because these observations were drawn from an artificial experiment, they should be verified in a “real world” study.

STATEMENT OF AUTHORSHIP

Category 1: a. Conception and design: D. B. Altshuler, C. A. Andrews, H. A. Parmar, S. E. Sullivan, and J. D. Trobe; b. Acquisition of data: D. B. Altshuler, C. A. Andrews, H. A. Parmar, S. E. Sullivan, and J. D. Trobe; c. Analysis and interpretation of data: D. B. Altshuler, C. A. Andrews, H. A. Parmar, S. E. Sullivan, and J. D. Trobe. Category 2: a. Drafting the manuscript: D. B. Altshuler and J. D. Trobe; b. Revising it for intellectual content: D. B. Altshuler and J. D. Trobe. Category 3: a. Final approval of the completed manuscript: D. B. Altshuler, C. A. Andrews, H. A. Parmar, S. E. Sullivan, and J. D. Trobe.

REFERENCES

1. Abele TA, Yetkin ZF, Raisanen JM, Mickey BE, Mendelsohn DB. Non-pituitary origin sellar tumours mimicking pituitary macroadenomas. Clin Radiol. 2012;67:821–827.
2. Chanson P, Brochier S. Non-functioning pituitary adenomas. J Endocrinol Invest. 2005;28:93–99.
3. Freda PU, Post KD Differential diagnosis of sellar masses. Endocrinol Metab Clin North Am. 1999;28, 81–117.
4. Rennert J, Doerfler A. Imaging of sellar and parasellar lesions. Clin Neurol Neurosurg. 2007;109:111–124.
5. Zee CS, Go JL, Kim PE, Mitchell D, Ahmadi J Imaging of the pituitary and parasellar region. Neurosurg Clin N Am. 2003;14, 55–80.
6. Matsumoto S, Hayase M, Imamura H, Oda Y, Kikuchi H, Katayama M, Ishihara T. A case of intrasellar meningioma mimicking pituitary adenoma. No Shinkei Geka. 2001;29:551–557.
7. Lach B, Reddy K, Sommer DD, Whitton AC, Baweja R. Amelanotic melanocytoma of the sella mimicking pituitary adenoma. Clin Neuropathol. 2017;36:283–287.
8. Locatelli M, Spagnoli D, Caroli M, Isalberti M, Branca V, Gaini SM, Lania A. A potential catastrophic trap: an unusually presenting sellar lesion. Eur J Neurol. 2008;15:98–101.
9. Gutenberg A, Larsen J, Lupi I, Rohde V, Caturegli P. A radiologic score to distinguish autoimmune hypophysitis from nonsecreting pituitary adenoma preoperatively. Am J Neuroradiol. 2009;30:1766–1772.
10. Furtado SV, Thakar S, Ghosal N, Hegde AS. Atypical presentation of pediatric mixed germ cell tumors in the sellar-suprasellar region. Neurol India. 2012;60:90–93.
11. Romano A, Chibbaro S, Marsella M, Ippolito S, Benericetti E. Carotid cavernous aneurysm presenting as pituitary apoplexy. J Clin Neurosci. 2006;13:476–479.
12. Haridas A, Ansari S, Afshar F. Chordoma presenting as pseudoprolactinoma. Br J Neurosurg. 2003;17:260–262.
13. Hanak BW, Zada G, Nayar VV, Thiex R, Du R, Day AL, Laws ER. Cerebral aneurysms with intrasellar extension: a systematic review of clinical, anatomical, and treatment characteristics. J Neurosurg. 2012;116:164–178.
14. Kwancharoen R, Blitz AM, Tavares F, Caturegli P, Gallia GL, Salvatori R. Clinical features of sellar and suprasellar meningiomas. Pituitary. 2014;17:342–348.
15. Juco J, Horvath E, Smyth H, Rotondo F, Kovacs K. Hemangiopericytoma of the sella mimicking pituitary adenoma: case report and review of the literature. Clin Neuropathol. 2007;26:288–293.
16. Nishioka H, Ito H, Haraoka J, Akada K. Immature teratoma originating from the pituitary gland: case report. Neurosurgery. 1999;44, 644–647; discussion 647-648 (1999).
17. Krayenbühl N, Heppner F, Yonekawa Y, Bernays RL. Intrasellar malignant peripheral nerve sheath tumor (MPNST). Acta Neurochir. 2007;149, 201–205; discussion 205-206 (2007).
18. Zhang J, Xu S, Liu Q, Li X, Jia D, Li G. Intrasellar and suprasellar schwannoma misdiagnosed as pituitary macroadenoma: a case report and Review of the Literature. World Neurosurg. 2016;96:612.e1–612.e7.
19. Terasaka T, Inagaki K, Otsuka F. Invasive intrasellar plasmacytoma mimicking pituitary adenoma. Intern Med Tokyo Jpn. 2016;55:1501–1502.
20. Sankhe S, Gandhi J, Shah NS, Khare S. Lymphocytic hypophysitis in nonpregnant middle-aged females mimicking pituitary adenoma. J Postgrad Med. 2016;62:50–51.
21. Bitterman P, Ariza A, Black RA, Allen WE, Lee SH. Multiple myeloma mimicking pituitary adenoma. Comput Radiol. 1986;10:201–205.
22. Inoue T, Takahashi N, Murakami K, Nishimura S, Kaimori M, Nishijima M. Osteochondroma of the sella turcica presenting with intratumoral hemorrhage. Neurol Med Chir (Tokyo). 2009;49:37–41.
23. Eom KS, See-Sung C, Kim JD, Kim JM, Kim TY. Primary hypothyroidism mimicking a pituitary macroadenoma: regression after thyroid hormone replacement therapy. Pediatr Radiol. 2009;39:164–167.
24. Kong X, Wu H, Ma W, Li Y, Yang Y, Xing B, Wei J, Yao Y, Gao J, Lian W, Xu Z, Dou W, Ren Z, Su C, Wang R. Schwannoma in sellar region mimics invasive pituitary macroadenoma: literature review with one case report. Medicine (Baltimore). 2016;95:e2931.
25. Glezer A, Paraiba DB, Bronstein MD. Rare sellar lesions. Endocrinol Metab Clin North Am. 2008;37:195–211.
26. Balasubramani Y, Ellul S, Kam A, McLean C, Malham G. Sinonasal inverted papilloma mimicking a pituitary macroadenoma. J Clin Neurosci. 2009;16:328–330.
27. Desai KI, Nadkarni TD, Goel A. Tuberculomas of the hypophysis cerebri: report of five cases. J Clin Neurosci. 2003;10:562–566.
Copyright © 2021 by North American Neuro-Ophthalmology Society