Pediatr Radiol DOI 10.1007/s00247-015-3308-x
ORIGINAL ARTICLE
Pediatric adrenocortical neoplasms: can imaging reliably discriminate adenomas from carcinomas? Kelsey A. Flynt & Jonathan R. Dillman & Matthew S. Davenport & Ethan A. Smith & Tobias Else & Peter J. Strouse & Elaine M. Caoili
Received: 9 October 2014 / Revised: 15 December 2014 / Accepted: 5 February 2015 # Springer-Verlag Berlin Heidelberg 2015
Abstract Background There is a paucity of literature describing and comparing the imaging features of adrenocortical adenomas and carcinomas in children and adolescents. Objective To document the CT and MRI features of adrenocortical neoplasms in a pediatric population and to determine whether imaging findings (other than metastatic disease) can distinguish adenomas from carcinomas. Materials and methods We searched institutional medical records to identify pediatric patients with adrenocortical neoplasms. Pre-treatment CT and MRI examinations were reviewed by two radiologists in consensus, and pertinent imaging findings were documented. We also recorded relevant histopathological, demographic, clinical follow-up and survival data. We used the Student’s t-test and Wilcoxon rank sum test to compare parametric and nonparametric continuous data, and the Fisher exact test to compare proportions. We used receiver operating characteristic (ROC) curve analyses to evaluate the diagnostic performances of tumor diameter and volume for discriminating carcinoma from adenoma. A P-value ≤0.05 was considered statistically significant. K. A. Flynt : J. R. Dillman (*) : E. A. Smith : P. J. Strouse Section of Pediatric Radiology, C. S. Mott Children’s Hospital, Department of Radiology, University of Michigan Health System, 1540 East Hospital Drive, Ann Arbor, MI 48109, USA e-mail: [email protected] M. S. Davenport : E. M. Caoili Division of Abdominal Imaging, Department of Radiology, University of Michigan Health System, Ann Arbor, MI, USA T. Else Division of Metabolism, Endocrinology and Diabetes, Department of Internal Medicine, University of Michigan Health System, Ann Arbor, MI, USA
Results Among the adrenocortical lesions, 9 were adenomas, 15 were carcinomas, and 1 was of uncertain malignant potential. There were no differences in mean age, gender or sidedness between adenomas and carcinomas. Carcinomas were significantly larger than adenomas based on mean estimated volume (581 ml, range 16–2,101 vs. 54 ml, range 3–197 ml; P-value= 0.003; ROC area under the curve=0.92) and mean maximum transverse plane diameter (9.9 cm, range 3.0–14.9 vs. 4.4 cm, range 1.9–8.2 cm; P-value=0.0001; ROC area under the curve=0.92). Carcinomas also were more heterogeneous than adenomas on post-contrast imaging (13/14 vs. 2/9; odds ratio [OR]=45.5; P-value=0.001). Six of 13 carcinomas and 1 of 8 adenomas contained calcification at CT (OR=6.0; P-value= 0.17). Seven of 15 children with carcinomas exhibited metastatic disease at diagnosis, and three had inferior vena cava invasion. Median survival for carcinomas was 27 months. Conclusion In our experience, pediatric adrenocortical carcinomas are larger, more heterogeneous, and more often calcified than adenomas, although there is overlap in their imaging appearances. Keywords Adrenocortical neoplasm . Carcinoma . Adenoma . Children . Computed tomography . Magnetic resonance imaging
Introduction Adrenocortical neoplasms in the pediatric population are rare, with an incidence of approximately three cases per 1 million; they are clinically and histologically different from those observed in the adult population and can be benign (adenomas), malignant (carcinomas), or of uncertain malignant potential [1, 2]. Although sometimes detected incidentally, these tumors most often present as a result of associated increased hormone production [3, 4]. Precocious puberty or virilization caused by
Pediatr Radiol
increased androgen production and Cushing syndrome caused by increased cortisol production are frequently reported presentations [1, 2, 5–7]. Pediatric adenomas and carcinomas have very different prognoses, with adenomas typically being cured by surgical resection. Despite aggressive surgical and medical treatment, carcinomas generally have a poor prognosis [6], with one study reporting a 5-year survival of 0% in children with stage 3 and stage 4 tumors [8]. A large study by Michalkiewicz et al. [6] documented an overall 5-year survival of 54.7% in children with adrenocortical carcinoma, although the 5-year event-free survival for stages 1 and 2 ranged from 38.1% in older children (13- to 20-year-olds) to 85.6% in very young children (0- to 3-year-olds). There is a paucity of literature describing and comparing the routine imaging features of pediatric adrenocortical neoplasms at CT and MRI. Based on available reports, pediatric adenomas appear different from those typically observed in the adult population (which are generally small, homogeneous adrenal gland nodules or masses) [3, 9]. Also, there is no validated CT or MRI method for assessing adrenal adenomas in children, unlike in the adult population, where a Hounsfield unit cut-off value (CT densitometry) is used to diagnose lipidrich adenomas [10] and contrast enhancement washout is used to discriminate lipid-poor adenomas from other adrenal lesions, such as metastases [11, 12]. Pretreatment knowledge of adenoma vs. carcinoma in the pediatric population could help direct surgical management (e.g., laparoscopic resection of adenomas vs. open radical resection of carcinomas) as well as guide the use of neoadjuvant chemotherapy, should this approach become standard of care. The purpose of our study was to document the routine CT and MRI features of adrenocortical neoplasms in our pediatric patient population. In addition, we sought to determine whether any imaging features (e.g., tumor diameter/volume, homogeneous vs. heterogeneous appearance, presence of calcification) other than the existence of metastatic disease can reliably distinguish pediatric adrenocortical adenomas from carcinomas.
Materials and methods This retrospective investigation was approved by our institutional review board and complies with the Health Insurance Portability and Accountability Act. The requirement for informed consent was waived. Institutional Departments of Radiology, Pathology and Oncology records were searched to identify all pediatric patients (younger than or equal to 18 years) with adrenocortical neoplasms (adenomas, carcinomas and lesions of uncertain malignant potential) between January 1995 and June 2014. We excluded children without pertinent pre-treatment (medical or surgical) CT, MRI or US imaging.
Relevant pre-treatment imaging examinations were reviewed by two radiologists in consensus (a third-year radiology resident and a fellowship-trained pediatric radiologist with 5 years of clinical experience). Pre-treatment CT and MRI examinations were reviewed (as available), while US imaging was only reviewed if no CT or MRI examination was available. CT and MRI examination parameters varied because imaging examinations were performed over a nearly 20-year period, and many studies were performed at outside institutions. The following pre-treatment imaging findings were documented: & & & & & & & & & &
Side of mass (right vs. left) Maximum dimensions of mass in transverse, anteroposterior and craniocaudad planes (cm) Estimated tumor volume (ml; using 0.52 x transverse x anteroposterior x craniocaudad maximum dimension measurements) Post-contrast enhancement pattern (predominantly homogeneous vs. predominantly heterogeneous) Presence of calcification, if CT imaging available (yes or no) Mean attenuation of mass, if non-contrast CT imaging available (Hounsfield unit [HU], using region-of-interest including central two-thirds of lesion) Contrast material washout percentage of mass, if non-contrast, portal venous and delayed phases of imaging available Presence of inferior vena cava invasion (yes or no) Presence of metastatic disease at time of initial presentation (yes or no) Location of metastatic disease at time of initial presentation, if present (e.g., regional lymph nodes, liver, lung).
Available CT, MRI and US examinations acquired after the initiation of medical or surgical treatment as well as pertinent clinical data were reviewed to determine whether children with malignancy that initially presented without metastatic disease eventually developed distant spread. Relevant histopathological records were reviewed for all children by a single investigator in order to classify pediatric adrenocortical lesions as adenomas, carcinomas or lesions of uncertain malignant potential. Electronic medical records were searched to obtain demographic information (e.g., age and gender) and clinical presentation (e.g., precocious puberty/virilization Cushing syndrome, primary aldosteronism, incidental finding). We documented length of follow-up (in months) from diagnosis to most recent clinical encounter in the medical record. For children with no medical records after Dec. 31, 2013, we searched the United States Social Security Administration death records to allow additional assessment of survival.
Pediatr Radiol
Statistical analysis Continuous data were summarized using means and ranges, while categorical data were summarized using counts and percentages. Continuous data were graphically explored using box-and-whisker plots, as appropriate. We used the Student’s t-test and Wilcoxon rank sum test to compare parametric and nonparametric continuous data, and the Fisher exact test to compare proportions. We performed receiver operating characteristic (ROC) curve analyses to evaluate the diagnostic performances of maximum transverse plane diameter and estimated tumor volume for discriminating adrenocortical carcinoma from adenoma. The Kaplan-Meier method was used to compare the median survival of children with adenomas compared to carcinomas. A P-value ≤0.05 was considered statistically significant for all inference testing. Statistical analyses were performed using GraphPad Prism 6 (GraphPad Software Inc., La Jolla, CA).
Results
Table 1 Demographic data and clinical presentations in children and adolescents with adrenocortical neoplasms
Gender Girls (n, %) Boys (n, %) Mean age (years) Overall (range) Girls (range) Boys (mean, range) Clinical presentations Precocious puberty/ virilization (n) Cushing syndrome (n) Primary aldosteronism (n) Abdominal pain Incidental/other (n)
Adenoma (n=9)
Carcinoma (n=15)
Uncertain malignant potential (n=1)
7 (78%) 2 (22%)
9 (60%) 6 (40%)
1
11.4 (2–17) 11.5 (0.17–18) 3 (no range) 12.0 (2–17) 14.1 (3–18) 9.5 (9–10) 7.5 (0.17–17) 3 (no range) 5
5
1
1 1
4 -
-
2
2 4
-
n number of subjects; - no subjects
Histopathological diagnosis All 25 pediatric patients with adrenocortical lesions and available CT, MRI or US imaging had correlative histopathological data. Of the adrenocortical lesions, 9 were adenomas, 15 were carcinomas, and 1 was of uncertain malignant potential. Demographic data and clinical presentations Demographic data and clinical presentations are presented in Table 1. There was no significant difference in the proportion of girls vs. boys when comparing adrenocortical adenomas to carcinomas in our patient population (P-value=0.66). Similarly, there was no significant difference in the mean ages of children with adenomas vs. carcinomas (11.4 vs. 11.5 years; P-value=0.92). Imaging features of pediatric adrenocortical neoplasms Adenoma Eight (89%) children with adenomas were imaged with CT, while a single (11%) child was imaged with MRI. Six adenomas were located on the right (66%); three were located on the left (33%). Mean maximum transverse plane dimension of adenomas was 4.4 cm (range 1.9–8.2 cm), while mean estimated tumor volume of adenomas was 54 ml (range 3–197 ml) (Fig. 1). Two (22%) adenomas appeared predominantly heterogeneous on post-contrast imaging, while only one (13%) adenoma contained calcification based on CT imaging (Fig. 2). Five (63%) of eight children with CT imaging had a
non-contrast phase; only one of these adenomas demonstrated a mean HU measurement value of less than 10, suggesting a lipid-rich adenoma. The other four adenomas with non-contrast CT imaging appeared lipid-poor based on CT densitometry, with HU measurements ranging from 30 to 46. Only two children had sufficient imaging available to calculate the washout percentage of contrast material from the mass, calculated to be 59% and 68%, respectively. As anticipated, no histologically confirmed adenoma showed evidence of inferior vena cava invasion or metastatic disease.
Carcinoma Twelve (80%) children with carcinomas were imaged with CT only, one (6.7%) child was imaged with MRI only, and one (6.7%) child was imaged with both CT and MRI. An additional single child (6.7%) was imaged with US but had no available CT or MR imaging; only demographic, tumor side and size, and follow-up/survival data were collected for this child. Eight carcinomas were located on the right (53%), while seven were located on the left (47%). Mean maximum transverse plane dimension of carcinomas was 9.9 cm (range 3.0– 14.9 cm), while mean carcinoma estimated tumor volume was 581.0 ml (range 16–2,101 ml) (Fig. 1); carcinoma volume could not be calculated for one lesion because the inferiormost portion of the mass was excluded from available CT imaging. Thirteen (93%) of 14 carcinomas with CT or MR imaging appeared predominantly heterogeneous on post-
Pediatr Radiol
Fig. 1 Size of carcinomas vs. adenomas. Dot plots show (a) maximum transverse plane diameter (cm), and (b) estimated volume (ml) of tumors for children with adrenocortical carcinomas vs. adenomas. For each
histological category, the middle horizontal line represents the mean, while the whiskers represent two standard deviations
contrast imaging (Figs. 3, 4 and 5). Six (46%) of 13 carcinomas with CT imaging contained calcification (Figs. 3 and 4). Three (23%) of 13 children with carcinomas and available CT imaging had a non-contrast phase; all three of these carcinomas demonstrated a mean Hounsfield unit measurement of greater than 10 (range 36–64 HU). No child with carcinoma had sufficient imaging available to calculate the washout percentage of contrast material from the mass. Seven (47%) of 15 children with carcinomas showed evidence of metastatic disease at the time of initial diagnosis. Metastatic disease was noted to involve the following sites: lung (n=5), liver (n=4), regional lymph nodes (n=2) and peritoneum (n=1) (Fig. 4). Three children exhibited delayed presentation of metastatic disease, all experiencing local retroperitoneal recurrences as well as pulmonary metastases or
metastatic disease to regional lymph nodes. Three children had evidence of inferior vena cava invasion at the time of initial diagnosis (Fig. 5); all three of these children had local retroperitoneal recurrences as well as evidence of pulmonary metastases (two at the time of diagnosis and one during therapy). There was no significant difference in the size of carcinomas presenting with vs. without metastatic disease (448.1 ml vs. 758.2 ml; P-value=0.3).
Fig. 2 Adrenocortical adenoma. Axial CT in a 16-year-old girl with an incidentally detected large right adrenocortical adenoma, which demonstrated central degeneration and hemorrhage at histology (not shown). The mass (arrows) heterogeneously enhances and contains small amounts of calcification (arrowhead)
Fig. 3 Adrenocortical carcinoma in a 3-year-old girl with precocious puberty. Coronal post-contrast CT image shows a large, heterogeneous, partly calcified (arrowhead) mass (arrows) arising from the right adrenal gland. The mass exerts substantial mass effect upon the right kidney, liver and inferior vena cava
Pediatr Radiol
Fig. 6 MRI of an adenoma in a 17-year-old girl presenting with virilization. Coronal T2-weighted fast spin-echo fat-saturated image shows a large right heterogeneous mass (arrows) arising from the right adrenal gland, confirmed at histology (not shown) to be an adenoma. This mass was the largest adenoma in our study
Fig. 4 Adrenocortical carcinoma in a 17-year-old girl with Cushing syndrome. Coronal post-contrast CT image shows a large, heterogeneous, partly calcified (arrowhead) mass (white arrows) arising from the left adrenal gland. The mass exerts substantial mass effect upon the left kidney. Low-attenuation liver masses (black arrows) are from metastatic carcinoma
Adenomas vs. carcinomas There were no significant differences in mean age of presentation, gender or sidedness (right vs. left) between adrenocortical adenomas and carcinomas. Although of variable size, carcinomas were significantly larger than adenomas based on mean estimated tumor volume (581 ml vs. 54 ml; P-value=0.003) and mean maximum transverse plane diameter (9.9 cm vs. 4.4 cm; P-value=0.0001) (Figs. 1, 6, 7, 8 and 9). Carcinomas also were significantly more heterogeneous than adenomas (13/14 vs. 2/9; P-value=0.001; OR=45.5; 95% confidence interval [CI], 3.5–595.1). Six of 13 carcinomas and 1 of 8 adenomas contained calcification on CT (P-value=0.17; OR=6.0; 95% CI, 0.57–63.7).
Fig. 5 Adrenocortical carcinoma in a 6-year-old boy with precocious puberty. a Axial postcontrast CT image shows a large right suprarenal mass (arrows). b Coronal post-contrast CT image shows tumor filling the inferior vena cava (arrows) and extending into the right atrium
ROC curves were created to assess the diagnostic performances of estimated tumor volume and maximum transverse plane diameter for discriminating adrenocortical carcinoma from adenoma. Areas under the curves were both 0.92, respectively (Fig. 10). An estimated tumor volume cut-off of 212.5 ml provided 79% sensitivity and 100% specificity for discriminating carcinoma from adenoma, while a maximum transverse plane diameter cut-off of 8.5 cm provided 80% sensitivity and 100% specificity.
Adrenocortical lesion of uncertain malignant potential Our only adrenocortical lesion of uncertain malignant potential based on histopathological analysis arose in a 3-year-old boy with Li-Fraumeni syndrome (p53 tumor suppressor gene mutation) who presented with precocious puberty. His predominantly homogeneous tumor had a maximum transverse plane diameter of 2.7 cm and an estimated tumor volume of 6.3 ml. No metastatic disease was identified at initial presentation or upon 43 months of follow-up.
Pediatr Radiol
Fig. 7 CT of an adenoma in a 2-year-old girl presenting with precocious puberty. Axial post-contrast CT image shows a round mass (arrow) arising from the left adrenal gland, confirmed to be an adenoma. A portion of the left adrenal gland is visible. This child has much more visceral and subcutaneous fat than expected for age because of hypercortisolism
Subject follow-up
Fig. 9 CT image of adrenocortical carcinoma in an 18-year-old woman. Axial post-contrast image shows an incidentally detected left adrenocortical carcinoma (arrows) anterior to the left kidney upper pole. The mass, the smallest carcinoma in our study, appears homogeneously solid and contains tiny calcifications (arrowhead)
three children were lost to follow-up after 12, 25 and 39 months, respectively. Median survival in our patient population was 27 months.
Adenoma All nine adrenocortical adenomas were surgically resected because of hormonal activity or concern about potential malignancy. Median follow-up for adenomas was 29 months (range 1–192 months). No child with an adenoma experienced local or distant recurrence, based on available medical records.
Carcinoma Median follow-up of adrenocortical carcinomas was 19 months (range 6–229 months). Twelve of 15 children had complete follow-up using the methods described above, while
Fig. 8 CT of an adrenocortical carcinoma in 13-year-old girl presenting with Cushing syndrome. Axial post-contrast CT image shows a very large, heterogeneous mass (arrows) arising from the left adrenal gland, confirmed to be adrenocortical carcinoma. The mass exerts substantial mass effect upon the left kidney, bowel and mesentery
Discussion Based on our study population, there was no significant demographic difference between children and adolescents with adrenocortical adenomas and carcinomas. The mean age of children with adenomas was 11.4 years compared to 11.5 years for those with carcinomas. This result differs from other published studies where pediatric adrenocortical lesions were found to occur more often in younger children [1, 2]. Interestingly, one of our adrenocortical carcinomas was likely congenital, coming to attention as a palpable mass at 2 months of age. Congenital adrenocortical carcinomas and adenomas are very rare but have been described [13]. Both adenomas and carcinomas were more common in girls, an observation that has been described by others [1, 5–8]. The majority of adrenocortical tumors in our study were hormonally active, often causing precocious puberty or virilization or Cushing syndrome (three children, including two who presented prior to 2000, did not have available laboratory data). A small but substantial number of adenomas and carcinomas presented with abdominal pain or were incidentally detected, with six of eight of these tumors shown to be hormone-producing based upon laboratory evaluation. Based on our experience, aldosterone-producing tumors and associated Conn syndrome (hypertension and hypokalemia) is probably rare in the pediatric population; feminization in boys from estrogen production was not observed but has been reported [1]. At CT and MRI, several differences between adenomas and carcinomas were observed. First, on average, carcinomas were
Pediatr Radiol
Fig. 10 Assessment of size to discriminate adrenocortical carcinoma from adenoma. Receiver operating characteristic (ROC) curves used to assess the diagnostic performance of (a) estimated tumor volume (ml),
and (b) maximum transverse plane diameter (cm) for discriminating adrenocortical carcinoma from adenoma. Areas under the curves are 0.92 and 0.92, respectively, suggesting good diagnostic accuracy
much larger than adenomas based on maximum transverse plane diameter (P-value=0.0001) and estimated volume (P= 0.003) of tumor. Both of these imaging features allowed similar discrimination between adenoma and carcinoma based on ROC analysis, with respectable areas under the curve of 0.92. Two adenomas were quite large (approximately 8 cm in maximum diameter), suggesting that a lower cut-off value for maximum transverse plane diameter might falsely classify a small number of adenomas as carcinomas. A cut-off maximum transverse plane diameter of 8.5 cm and cut-off estimated tumor volume of 212.5 ml were both 100% specific for carcinoma based on our data. Not surprisingly, in three children adrenocortical carcinomas were diagnosed with diameters and volumes that clearly overlapped with the majority of adenomas and which did not have imaging evidence of metastatic disease. It is such malignant lesions that pediatric radiologists cannot reliably distinguish from benign adenomas based on standard CT and MRI techniques. Other imaging features were also seen more commonly in carcinomas compared to adenomas, including a predominantly heterogeneous pattern of enhancement on post-contrast imaging (P-value = 0.001) and presence of calcification (P-value=0.17). All but one carcinoma heterogeneously enhanced (commonly with a central stellate appearance), while only two adenomas shared this imaging feature. The magnitude of association between adrenocortical lesion heterogeneity and the histological diagnosis of carcinoma was striking, with an odds ratio of 45.5. That is, the odds of an adrenocortical lesion being carcinoma (vs. adenoma) are 45.5 times greater when post-contrast enhancement is heterogeneous. Although the 95% confidence interval accompanying this odds ratio is very wide because of our relatively small sample
population, the effect size suggests a very strong relationship. We want to acknowledge, however, that adrenocortical lesion heterogeneity might be impacted by the timing of imaging as well as the rapidity of injection of the intravenous contrast material. Six of 13 carcinomas also contained calcification, while only a single adenoma shared this imaging feature. Although this difference in the proportion of calcified carcinomas vs. adenomas was not statistically significant, the lack of significance again may be a result of our relatively small sample population and insufficient power, because a large magnitude of effect was observed (OR=6.0; 95% CI, 0.57–63.7). It is noteworthy that intralesional calcifications were detected in numerous children despite the presence of intravenous contrast material, highlighting the fact that non-contrast imaging is often unnecessary to detect calcification. Our results suggest that adrenocortical neoplasms that both heterogeneously enhance and contain calcification are at substantially increased odds of being carcinoma as opposed to adenoma, and such lesions greater than 8.5 cm are almost certainly malignant. The ability to discriminate adrenocortical adenomas from carcinomas prior to treatment could have benefits. First, although hormone-producing suspected adrenocortical masses in children are almost always removed, the surgical approach could be personalized to improve patient outcomes and decrease morbidity and mortality. Current preference is to laparoscopically resect adenomas, while carcinomas require radical open resection to remove the tumor en bloc [14]. If a lesion could be shown to be benign, a laparoscopic approach would be much preferred because of its lower surgical morbidity, lesser cosmetic impact, more rapid postoperative recovery with shorter hospitalization, and lower cost. Outcomes in children with carcinomas are strongly impacted by the
Pediatr Radiol
ability to achieve complete surgical resection of the tumor; thus open surgery is preferred in this setting [5, 14–16]. A recent study by Miller et al. [15] concluded that “open adrenalectomy is superior to laparoscopic adrenalectomy for adrenocortical carcinoma based on completeness of resection, site and timing of initial tumor recurrence, and survival in stage II patients.” Second, distinguishing adenomas from carcinomas could help direct chemotherapy if neoadjuvant treatment is considered prior to definitive surgery. A recent German nonrandomized, single-arm study indicated that neoadjuvant chemotherapy may play a role in certain children with adrenocortical carcinomas, such as those that are unresectable at the time of diagnosis; however, in general, surgical resection of the tumor should be performed as soon as possible after diagnosis [17]. Almost half of our subjects with carcinomas presented with metastatic disease, most often to the liver and lung. These are the two most common sites of metastases described in the existing literature as well [1, 4]. Surprisingly, no relationship between size of the primary tumor and the presence of distant metastatic disease was identified. Although only three of 15 children with carcinoma demonstrated inferior vena cava invasion, all of these children experienced pulmonary metastatic disease as well as local retroperitoneal recurrences. In some instances, tumor thrombus extends into the heart [18]. Based on available follow-up data, the median survival of children with carcinoma was 27 months. At the time of publication 8 of 15 children had died, while 4 were alive and 3 were lost to follow-up after varying lengths of follow-up. In the end, and based in part on prior studies, overall survival in children with adrenocortical neoplasms likely relates to a variety of factors, including patient age, histological features of the lesion, tumor biology/genetics, and tumor size (weight) [6, 19]. Remarkably, all adrenocortical lesions confirmed to be adenomas by histopathology behaved in a benign manner suggesting a correct diagnosis, despite the fact that distinguishing benign from malignant adrenocortical neoplasms in the pediatric population has been historically difficult [1, 2]. This suggests that the ability of pathologists to correctly discriminate adrenocortical adenomas from carcinomas may have substantially improved in recent years based on better diagnostic criteria. All children with adenomas were alive at the time of this report based on available records. Our study has limitations. First, we have a relatively small number of pediatric adrenocortical lesions (n=25), a fact that restricts the conclusions we can make from our data. However, our study is still one of the largest investigations documenting and comparing the imaging features of adenomas and carcinomas in children and adolescents given the rarity of these neoplasms. Second, many more children had undergone CT evaluation than MRI evaluation. Thus, we have insufficient data from which to draw conclusions regarding specific MRI findings that might discriminate adenomas from
carcinomas, such as apparent diffusion coefficient value or loss of signal on T1-weighted gradient recalled echo out-ofphase imaging. Third, we are unable to adequately assess tumor washout of contrast material based on multi-phase CT, a technique that is fundamental to characterizing adrenocortical masses (in particular, adenomas) in adults. Necessary imaging data for washout percentage calculation was available for only two children because pediatric CT examinations performed at our institution are most often a single portal venous phase in order to adhere to “Image Gently” and “as low as reasonably achievable” (ALARA) principles. Finally, in a small number of children (three) with carcinoma we have incomplete follow-up, although Social Security Administration data suggest that these children are still alive.
Conclusion Pediatric adrenocortical carcinomas are larger and more heterogeneous than adenomas and commonly contain calcification. Using the tumor diameter and estimated volume cut-off values mentioned above, CT and MRI discriminate carcinomas from adenomas with a good degree of accuracy (areas under the curve=0.92 and 0.92, respectively); however very small carcinomas and very large adenomas occur, making complete discrimination of these two histological entities impossible based on conventional imaging findings. Further research is needed to determine whether more advanced imaging techniques, such as dynamic contrast-enhanced, chemicalshift and diffusion-weighted MRI, can allow definitive discrimination of adenomas from carcinomas in children and adolescents.
Conflicts of interest None
References 1. Agrons GA, Lonergan GJ, Dickey GE et al (1999) Adrenocortical neoplasms in children: radiologic-pathologic correlation. Radiographics 19:989–1008 2. Wieneke JA, Thompson LD, Heffess CS (2003) Adrenal cortical neoplasms in the pediatric population: a clinicopathologic and immunophenotypic analysis of 83 patients. Am J Surg Pathol 27: 867–881 3. Daneman A, Chan HS, Martin J (1983) Adrenal carcinoma and adenoma in children: a review of 17 patients. Pediatr Radiol 13:11–18 4. Ribeiro J, Ribeiro RC, Fletcher BD (2000) Imaging findings in pediatric adrenocortical carcinoma. Pediatr Radiol 30:45–51 5. Hubertus J, Boxberger N, Redlich A et al (2012) Surgical aspects in the treatment of adrenocortical carcinomas in children: data of the GPOH-MET 97 trial. Klin Padiatr 224:143–147 6. Michalkiewicz E, Sandrini R, Figueiredo B et al (2004) Clinical and outcome characteristics of children with adrenocortical tumors: a
Pediatr Radiol report from the International Pediatric Adrenocortical Tumor Registry. J Clin Oncol 22:838–845 7. Neblett WW, Frexes-Steed M, Scott HW Jr (1987) Experience with adrenocortical neoplasms in childhood. Am Surg 53:117–125 8. Hanna AM, Pham TH, Askegard-Giesmann JR et al (2008) Outcome of adrenocortical tumors in children. J Pediatr Surg 43:843–849 9. Hanson JA, Weber A, Reznek RH et al (1996) Magnetic resonance imaging of adrenocortical adenomas in childhood: correlation with computed tomography and ultrasound. Pediatr Radiol 26:794–799 10. Israel GM, Korobkin M, Wang C et al (2004) Comparison of unenhanced CT and chemical shift MRI in evaluating lipid-rich adrenal adenomas. AJR Am J Roentgenol 83:215–219 11. Caoili EM, Korobkin M, Francis IR et al (2002) Adrenal masses: characterization with combined unenhanced and delayed enhanced CT. Radiology 222:629–633 12. Korobkin M, Brodeur FJ, Francis IR et al (1998) CT time-attenuation washout curves of adrenal adenomas and nonadenomas. AJR Am J Roentgenol 170:747–752 13. Sarwar ZU, Ward VL, Mooney DP et al (2004) Congenital adrenocortical adenoma: case report and review of literature. Pediatr Radiol 34:991–994
14. Miller BS, Doherty GM (2014) Surgical management of adrenocortical tumours. Nat Rev Endocrinol 10:282–292 15. Miller BS, Gauger PG, Hammer GD et al (2012) Resection of adrenocortical carcinoma is less complete and local recurrence occurs sooner and more often after laparoscopic adrenalectomy than after open adrenalectomy. Surgery 152:1150–1157 16. Else T, Williams AR, Sabolch A et al (2014) Adjuvant therapies and patient and tumor characteristics associated with survival of adult patients with adrenocortical carcinoma. J Clin Endocrinol Metab 99:455–461 17. Redlich A, Boxberger N, Strugala D et al (2012) Systemic treatment of adrenocortical carcinoma in children: data from the German GPOH-MET 97 trial. Klin Padiatr 224:366–371 18. Godine LB, Berdon WE, Brasch RC et al (1990) Adrenocortical carcinoma with extension into inferior vena cava and right atrium: report of 3 cases in children. Pediatr Radiol 20:166–168, discussion 169 19. Bugg MF, Ribeiro RC, Roberson PK et al (1994) Correlation of pathologic features with clinical outcome in pediatric adrenocortical neoplasia. A study of a Brazilian population. Brazilian Group for Treatment of Childhood Adrenocortical Tumors. Am J Clin Pathol 101:625–629
ORIGINAL ARTICLE
Pediatric adrenocortical neoplasms: can imaging reliably discriminate adenomas from carcinomas? Kelsey A. Flynt & Jonathan R. Dillman & Matthew S. Davenport & Ethan A. Smith & Tobias Else & Peter J. Strouse & Elaine M. Caoili
Received: 9 October 2014 / Revised: 15 December 2014 / Accepted: 5 February 2015 # Springer-Verlag Berlin Heidelberg 2015
Abstract Background There is a paucity of literature describing and comparing the imaging features of adrenocortical adenomas and carcinomas in children and adolescents. Objective To document the CT and MRI features of adrenocortical neoplasms in a pediatric population and to determine whether imaging findings (other than metastatic disease) can distinguish adenomas from carcinomas. Materials and methods We searched institutional medical records to identify pediatric patients with adrenocortical neoplasms. Pre-treatment CT and MRI examinations were reviewed by two radiologists in consensus, and pertinent imaging findings were documented. We also recorded relevant histopathological, demographic, clinical follow-up and survival data. We used the Student’s t-test and Wilcoxon rank sum test to compare parametric and nonparametric continuous data, and the Fisher exact test to compare proportions. We used receiver operating characteristic (ROC) curve analyses to evaluate the diagnostic performances of tumor diameter and volume for discriminating carcinoma from adenoma. A P-value ≤0.05 was considered statistically significant. K. A. Flynt : J. R. Dillman (*) : E. A. Smith : P. J. Strouse Section of Pediatric Radiology, C. S. Mott Children’s Hospital, Department of Radiology, University of Michigan Health System, 1540 East Hospital Drive, Ann Arbor, MI 48109, USA e-mail: [email protected] M. S. Davenport : E. M. Caoili Division of Abdominal Imaging, Department of Radiology, University of Michigan Health System, Ann Arbor, MI, USA T. Else Division of Metabolism, Endocrinology and Diabetes, Department of Internal Medicine, University of Michigan Health System, Ann Arbor, MI, USA
Results Among the adrenocortical lesions, 9 were adenomas, 15 were carcinomas, and 1 was of uncertain malignant potential. There were no differences in mean age, gender or sidedness between adenomas and carcinomas. Carcinomas were significantly larger than adenomas based on mean estimated volume (581 ml, range 16–2,101 vs. 54 ml, range 3–197 ml; P-value= 0.003; ROC area under the curve=0.92) and mean maximum transverse plane diameter (9.9 cm, range 3.0–14.9 vs. 4.4 cm, range 1.9–8.2 cm; P-value=0.0001; ROC area under the curve=0.92). Carcinomas also were more heterogeneous than adenomas on post-contrast imaging (13/14 vs. 2/9; odds ratio [OR]=45.5; P-value=0.001). Six of 13 carcinomas and 1 of 8 adenomas contained calcification at CT (OR=6.0; P-value= 0.17). Seven of 15 children with carcinomas exhibited metastatic disease at diagnosis, and three had inferior vena cava invasion. Median survival for carcinomas was 27 months. Conclusion In our experience, pediatric adrenocortical carcinomas are larger, more heterogeneous, and more often calcified than adenomas, although there is overlap in their imaging appearances. Keywords Adrenocortical neoplasm . Carcinoma . Adenoma . Children . Computed tomography . Magnetic resonance imaging
Introduction Adrenocortical neoplasms in the pediatric population are rare, with an incidence of approximately three cases per 1 million; they are clinically and histologically different from those observed in the adult population and can be benign (adenomas), malignant (carcinomas), or of uncertain malignant potential [1, 2]. Although sometimes detected incidentally, these tumors most often present as a result of associated increased hormone production [3, 4]. Precocious puberty or virilization caused by
Pediatr Radiol
increased androgen production and Cushing syndrome caused by increased cortisol production are frequently reported presentations [1, 2, 5–7]. Pediatric adenomas and carcinomas have very different prognoses, with adenomas typically being cured by surgical resection. Despite aggressive surgical and medical treatment, carcinomas generally have a poor prognosis [6], with one study reporting a 5-year survival of 0% in children with stage 3 and stage 4 tumors [8]. A large study by Michalkiewicz et al. [6] documented an overall 5-year survival of 54.7% in children with adrenocortical carcinoma, although the 5-year event-free survival for stages 1 and 2 ranged from 38.1% in older children (13- to 20-year-olds) to 85.6% in very young children (0- to 3-year-olds). There is a paucity of literature describing and comparing the routine imaging features of pediatric adrenocortical neoplasms at CT and MRI. Based on available reports, pediatric adenomas appear different from those typically observed in the adult population (which are generally small, homogeneous adrenal gland nodules or masses) [3, 9]. Also, there is no validated CT or MRI method for assessing adrenal adenomas in children, unlike in the adult population, where a Hounsfield unit cut-off value (CT densitometry) is used to diagnose lipidrich adenomas [10] and contrast enhancement washout is used to discriminate lipid-poor adenomas from other adrenal lesions, such as metastases [11, 12]. Pretreatment knowledge of adenoma vs. carcinoma in the pediatric population could help direct surgical management (e.g., laparoscopic resection of adenomas vs. open radical resection of carcinomas) as well as guide the use of neoadjuvant chemotherapy, should this approach become standard of care. The purpose of our study was to document the routine CT and MRI features of adrenocortical neoplasms in our pediatric patient population. In addition, we sought to determine whether any imaging features (e.g., tumor diameter/volume, homogeneous vs. heterogeneous appearance, presence of calcification) other than the existence of metastatic disease can reliably distinguish pediatric adrenocortical adenomas from carcinomas.
Materials and methods This retrospective investigation was approved by our institutional review board and complies with the Health Insurance Portability and Accountability Act. The requirement for informed consent was waived. Institutional Departments of Radiology, Pathology and Oncology records were searched to identify all pediatric patients (younger than or equal to 18 years) with adrenocortical neoplasms (adenomas, carcinomas and lesions of uncertain malignant potential) between January 1995 and June 2014. We excluded children without pertinent pre-treatment (medical or surgical) CT, MRI or US imaging.
Relevant pre-treatment imaging examinations were reviewed by two radiologists in consensus (a third-year radiology resident and a fellowship-trained pediatric radiologist with 5 years of clinical experience). Pre-treatment CT and MRI examinations were reviewed (as available), while US imaging was only reviewed if no CT or MRI examination was available. CT and MRI examination parameters varied because imaging examinations were performed over a nearly 20-year period, and many studies were performed at outside institutions. The following pre-treatment imaging findings were documented: & & & & & & & & & &
Side of mass (right vs. left) Maximum dimensions of mass in transverse, anteroposterior and craniocaudad planes (cm) Estimated tumor volume (ml; using 0.52 x transverse x anteroposterior x craniocaudad maximum dimension measurements) Post-contrast enhancement pattern (predominantly homogeneous vs. predominantly heterogeneous) Presence of calcification, if CT imaging available (yes or no) Mean attenuation of mass, if non-contrast CT imaging available (Hounsfield unit [HU], using region-of-interest including central two-thirds of lesion) Contrast material washout percentage of mass, if non-contrast, portal venous and delayed phases of imaging available Presence of inferior vena cava invasion (yes or no) Presence of metastatic disease at time of initial presentation (yes or no) Location of metastatic disease at time of initial presentation, if present (e.g., regional lymph nodes, liver, lung).
Available CT, MRI and US examinations acquired after the initiation of medical or surgical treatment as well as pertinent clinical data were reviewed to determine whether children with malignancy that initially presented without metastatic disease eventually developed distant spread. Relevant histopathological records were reviewed for all children by a single investigator in order to classify pediatric adrenocortical lesions as adenomas, carcinomas or lesions of uncertain malignant potential. Electronic medical records were searched to obtain demographic information (e.g., age and gender) and clinical presentation (e.g., precocious puberty/virilization Cushing syndrome, primary aldosteronism, incidental finding). We documented length of follow-up (in months) from diagnosis to most recent clinical encounter in the medical record. For children with no medical records after Dec. 31, 2013, we searched the United States Social Security Administration death records to allow additional assessment of survival.
Pediatr Radiol
Statistical analysis Continuous data were summarized using means and ranges, while categorical data were summarized using counts and percentages. Continuous data were graphically explored using box-and-whisker plots, as appropriate. We used the Student’s t-test and Wilcoxon rank sum test to compare parametric and nonparametric continuous data, and the Fisher exact test to compare proportions. We performed receiver operating characteristic (ROC) curve analyses to evaluate the diagnostic performances of maximum transverse plane diameter and estimated tumor volume for discriminating adrenocortical carcinoma from adenoma. The Kaplan-Meier method was used to compare the median survival of children with adenomas compared to carcinomas. A P-value ≤0.05 was considered statistically significant for all inference testing. Statistical analyses were performed using GraphPad Prism 6 (GraphPad Software Inc., La Jolla, CA).
Results
Table 1 Demographic data and clinical presentations in children and adolescents with adrenocortical neoplasms
Gender Girls (n, %) Boys (n, %) Mean age (years) Overall (range) Girls (range) Boys (mean, range) Clinical presentations Precocious puberty/ virilization (n) Cushing syndrome (n) Primary aldosteronism (n) Abdominal pain Incidental/other (n)
Adenoma (n=9)
Carcinoma (n=15)
Uncertain malignant potential (n=1)
7 (78%) 2 (22%)
9 (60%) 6 (40%)
1
11.4 (2–17) 11.5 (0.17–18) 3 (no range) 12.0 (2–17) 14.1 (3–18) 9.5 (9–10) 7.5 (0.17–17) 3 (no range) 5
5
1
1 1
4 -
-
2
2 4
-
n number of subjects; - no subjects
Histopathological diagnosis All 25 pediatric patients with adrenocortical lesions and available CT, MRI or US imaging had correlative histopathological data. Of the adrenocortical lesions, 9 were adenomas, 15 were carcinomas, and 1 was of uncertain malignant potential. Demographic data and clinical presentations Demographic data and clinical presentations are presented in Table 1. There was no significant difference in the proportion of girls vs. boys when comparing adrenocortical adenomas to carcinomas in our patient population (P-value=0.66). Similarly, there was no significant difference in the mean ages of children with adenomas vs. carcinomas (11.4 vs. 11.5 years; P-value=0.92). Imaging features of pediatric adrenocortical neoplasms Adenoma Eight (89%) children with adenomas were imaged with CT, while a single (11%) child was imaged with MRI. Six adenomas were located on the right (66%); three were located on the left (33%). Mean maximum transverse plane dimension of adenomas was 4.4 cm (range 1.9–8.2 cm), while mean estimated tumor volume of adenomas was 54 ml (range 3–197 ml) (Fig. 1). Two (22%) adenomas appeared predominantly heterogeneous on post-contrast imaging, while only one (13%) adenoma contained calcification based on CT imaging (Fig. 2). Five (63%) of eight children with CT imaging had a
non-contrast phase; only one of these adenomas demonstrated a mean HU measurement value of less than 10, suggesting a lipid-rich adenoma. The other four adenomas with non-contrast CT imaging appeared lipid-poor based on CT densitometry, with HU measurements ranging from 30 to 46. Only two children had sufficient imaging available to calculate the washout percentage of contrast material from the mass, calculated to be 59% and 68%, respectively. As anticipated, no histologically confirmed adenoma showed evidence of inferior vena cava invasion or metastatic disease.
Carcinoma Twelve (80%) children with carcinomas were imaged with CT only, one (6.7%) child was imaged with MRI only, and one (6.7%) child was imaged with both CT and MRI. An additional single child (6.7%) was imaged with US but had no available CT or MR imaging; only demographic, tumor side and size, and follow-up/survival data were collected for this child. Eight carcinomas were located on the right (53%), while seven were located on the left (47%). Mean maximum transverse plane dimension of carcinomas was 9.9 cm (range 3.0– 14.9 cm), while mean carcinoma estimated tumor volume was 581.0 ml (range 16–2,101 ml) (Fig. 1); carcinoma volume could not be calculated for one lesion because the inferiormost portion of the mass was excluded from available CT imaging. Thirteen (93%) of 14 carcinomas with CT or MR imaging appeared predominantly heterogeneous on post-
Pediatr Radiol
Fig. 1 Size of carcinomas vs. adenomas. Dot plots show (a) maximum transverse plane diameter (cm), and (b) estimated volume (ml) of tumors for children with adrenocortical carcinomas vs. adenomas. For each
histological category, the middle horizontal line represents the mean, while the whiskers represent two standard deviations
contrast imaging (Figs. 3, 4 and 5). Six (46%) of 13 carcinomas with CT imaging contained calcification (Figs. 3 and 4). Three (23%) of 13 children with carcinomas and available CT imaging had a non-contrast phase; all three of these carcinomas demonstrated a mean Hounsfield unit measurement of greater than 10 (range 36–64 HU). No child with carcinoma had sufficient imaging available to calculate the washout percentage of contrast material from the mass. Seven (47%) of 15 children with carcinomas showed evidence of metastatic disease at the time of initial diagnosis. Metastatic disease was noted to involve the following sites: lung (n=5), liver (n=4), regional lymph nodes (n=2) and peritoneum (n=1) (Fig. 4). Three children exhibited delayed presentation of metastatic disease, all experiencing local retroperitoneal recurrences as well as pulmonary metastases or
metastatic disease to regional lymph nodes. Three children had evidence of inferior vena cava invasion at the time of initial diagnosis (Fig. 5); all three of these children had local retroperitoneal recurrences as well as evidence of pulmonary metastases (two at the time of diagnosis and one during therapy). There was no significant difference in the size of carcinomas presenting with vs. without metastatic disease (448.1 ml vs. 758.2 ml; P-value=0.3).
Fig. 2 Adrenocortical adenoma. Axial CT in a 16-year-old girl with an incidentally detected large right adrenocortical adenoma, which demonstrated central degeneration and hemorrhage at histology (not shown). The mass (arrows) heterogeneously enhances and contains small amounts of calcification (arrowhead)
Fig. 3 Adrenocortical carcinoma in a 3-year-old girl with precocious puberty. Coronal post-contrast CT image shows a large, heterogeneous, partly calcified (arrowhead) mass (arrows) arising from the right adrenal gland. The mass exerts substantial mass effect upon the right kidney, liver and inferior vena cava
Pediatr Radiol
Fig. 6 MRI of an adenoma in a 17-year-old girl presenting with virilization. Coronal T2-weighted fast spin-echo fat-saturated image shows a large right heterogeneous mass (arrows) arising from the right adrenal gland, confirmed at histology (not shown) to be an adenoma. This mass was the largest adenoma in our study
Fig. 4 Adrenocortical carcinoma in a 17-year-old girl with Cushing syndrome. Coronal post-contrast CT image shows a large, heterogeneous, partly calcified (arrowhead) mass (white arrows) arising from the left adrenal gland. The mass exerts substantial mass effect upon the left kidney. Low-attenuation liver masses (black arrows) are from metastatic carcinoma
Adenomas vs. carcinomas There were no significant differences in mean age of presentation, gender or sidedness (right vs. left) between adrenocortical adenomas and carcinomas. Although of variable size, carcinomas were significantly larger than adenomas based on mean estimated tumor volume (581 ml vs. 54 ml; P-value=0.003) and mean maximum transverse plane diameter (9.9 cm vs. 4.4 cm; P-value=0.0001) (Figs. 1, 6, 7, 8 and 9). Carcinomas also were significantly more heterogeneous than adenomas (13/14 vs. 2/9; P-value=0.001; OR=45.5; 95% confidence interval [CI], 3.5–595.1). Six of 13 carcinomas and 1 of 8 adenomas contained calcification on CT (P-value=0.17; OR=6.0; 95% CI, 0.57–63.7).
Fig. 5 Adrenocortical carcinoma in a 6-year-old boy with precocious puberty. a Axial postcontrast CT image shows a large right suprarenal mass (arrows). b Coronal post-contrast CT image shows tumor filling the inferior vena cava (arrows) and extending into the right atrium
ROC curves were created to assess the diagnostic performances of estimated tumor volume and maximum transverse plane diameter for discriminating adrenocortical carcinoma from adenoma. Areas under the curves were both 0.92, respectively (Fig. 10). An estimated tumor volume cut-off of 212.5 ml provided 79% sensitivity and 100% specificity for discriminating carcinoma from adenoma, while a maximum transverse plane diameter cut-off of 8.5 cm provided 80% sensitivity and 100% specificity.
Adrenocortical lesion of uncertain malignant potential Our only adrenocortical lesion of uncertain malignant potential based on histopathological analysis arose in a 3-year-old boy with Li-Fraumeni syndrome (p53 tumor suppressor gene mutation) who presented with precocious puberty. His predominantly homogeneous tumor had a maximum transverse plane diameter of 2.7 cm and an estimated tumor volume of 6.3 ml. No metastatic disease was identified at initial presentation or upon 43 months of follow-up.
Pediatr Radiol
Fig. 7 CT of an adenoma in a 2-year-old girl presenting with precocious puberty. Axial post-contrast CT image shows a round mass (arrow) arising from the left adrenal gland, confirmed to be an adenoma. A portion of the left adrenal gland is visible. This child has much more visceral and subcutaneous fat than expected for age because of hypercortisolism
Subject follow-up
Fig. 9 CT image of adrenocortical carcinoma in an 18-year-old woman. Axial post-contrast image shows an incidentally detected left adrenocortical carcinoma (arrows) anterior to the left kidney upper pole. The mass, the smallest carcinoma in our study, appears homogeneously solid and contains tiny calcifications (arrowhead)
three children were lost to follow-up after 12, 25 and 39 months, respectively. Median survival in our patient population was 27 months.
Adenoma All nine adrenocortical adenomas were surgically resected because of hormonal activity or concern about potential malignancy. Median follow-up for adenomas was 29 months (range 1–192 months). No child with an adenoma experienced local or distant recurrence, based on available medical records.
Carcinoma Median follow-up of adrenocortical carcinomas was 19 months (range 6–229 months). Twelve of 15 children had complete follow-up using the methods described above, while
Fig. 8 CT of an adrenocortical carcinoma in 13-year-old girl presenting with Cushing syndrome. Axial post-contrast CT image shows a very large, heterogeneous mass (arrows) arising from the left adrenal gland, confirmed to be adrenocortical carcinoma. The mass exerts substantial mass effect upon the left kidney, bowel and mesentery
Discussion Based on our study population, there was no significant demographic difference between children and adolescents with adrenocortical adenomas and carcinomas. The mean age of children with adenomas was 11.4 years compared to 11.5 years for those with carcinomas. This result differs from other published studies where pediatric adrenocortical lesions were found to occur more often in younger children [1, 2]. Interestingly, one of our adrenocortical carcinomas was likely congenital, coming to attention as a palpable mass at 2 months of age. Congenital adrenocortical carcinomas and adenomas are very rare but have been described [13]. Both adenomas and carcinomas were more common in girls, an observation that has been described by others [1, 5–8]. The majority of adrenocortical tumors in our study were hormonally active, often causing precocious puberty or virilization or Cushing syndrome (three children, including two who presented prior to 2000, did not have available laboratory data). A small but substantial number of adenomas and carcinomas presented with abdominal pain or were incidentally detected, with six of eight of these tumors shown to be hormone-producing based upon laboratory evaluation. Based on our experience, aldosterone-producing tumors and associated Conn syndrome (hypertension and hypokalemia) is probably rare in the pediatric population; feminization in boys from estrogen production was not observed but has been reported [1]. At CT and MRI, several differences between adenomas and carcinomas were observed. First, on average, carcinomas were
Pediatr Radiol
Fig. 10 Assessment of size to discriminate adrenocortical carcinoma from adenoma. Receiver operating characteristic (ROC) curves used to assess the diagnostic performance of (a) estimated tumor volume (ml),
and (b) maximum transverse plane diameter (cm) for discriminating adrenocortical carcinoma from adenoma. Areas under the curves are 0.92 and 0.92, respectively, suggesting good diagnostic accuracy
much larger than adenomas based on maximum transverse plane diameter (P-value=0.0001) and estimated volume (P= 0.003) of tumor. Both of these imaging features allowed similar discrimination between adenoma and carcinoma based on ROC analysis, with respectable areas under the curve of 0.92. Two adenomas were quite large (approximately 8 cm in maximum diameter), suggesting that a lower cut-off value for maximum transverse plane diameter might falsely classify a small number of adenomas as carcinomas. A cut-off maximum transverse plane diameter of 8.5 cm and cut-off estimated tumor volume of 212.5 ml were both 100% specific for carcinoma based on our data. Not surprisingly, in three children adrenocortical carcinomas were diagnosed with diameters and volumes that clearly overlapped with the majority of adenomas and which did not have imaging evidence of metastatic disease. It is such malignant lesions that pediatric radiologists cannot reliably distinguish from benign adenomas based on standard CT and MRI techniques. Other imaging features were also seen more commonly in carcinomas compared to adenomas, including a predominantly heterogeneous pattern of enhancement on post-contrast imaging (P-value = 0.001) and presence of calcification (P-value=0.17). All but one carcinoma heterogeneously enhanced (commonly with a central stellate appearance), while only two adenomas shared this imaging feature. The magnitude of association between adrenocortical lesion heterogeneity and the histological diagnosis of carcinoma was striking, with an odds ratio of 45.5. That is, the odds of an adrenocortical lesion being carcinoma (vs. adenoma) are 45.5 times greater when post-contrast enhancement is heterogeneous. Although the 95% confidence interval accompanying this odds ratio is very wide because of our relatively small sample
population, the effect size suggests a very strong relationship. We want to acknowledge, however, that adrenocortical lesion heterogeneity might be impacted by the timing of imaging as well as the rapidity of injection of the intravenous contrast material. Six of 13 carcinomas also contained calcification, while only a single adenoma shared this imaging feature. Although this difference in the proportion of calcified carcinomas vs. adenomas was not statistically significant, the lack of significance again may be a result of our relatively small sample population and insufficient power, because a large magnitude of effect was observed (OR=6.0; 95% CI, 0.57–63.7). It is noteworthy that intralesional calcifications were detected in numerous children despite the presence of intravenous contrast material, highlighting the fact that non-contrast imaging is often unnecessary to detect calcification. Our results suggest that adrenocortical neoplasms that both heterogeneously enhance and contain calcification are at substantially increased odds of being carcinoma as opposed to adenoma, and such lesions greater than 8.5 cm are almost certainly malignant. The ability to discriminate adrenocortical adenomas from carcinomas prior to treatment could have benefits. First, although hormone-producing suspected adrenocortical masses in children are almost always removed, the surgical approach could be personalized to improve patient outcomes and decrease morbidity and mortality. Current preference is to laparoscopically resect adenomas, while carcinomas require radical open resection to remove the tumor en bloc [14]. If a lesion could be shown to be benign, a laparoscopic approach would be much preferred because of its lower surgical morbidity, lesser cosmetic impact, more rapid postoperative recovery with shorter hospitalization, and lower cost. Outcomes in children with carcinomas are strongly impacted by the
Pediatr Radiol
ability to achieve complete surgical resection of the tumor; thus open surgery is preferred in this setting [5, 14–16]. A recent study by Miller et al. [15] concluded that “open adrenalectomy is superior to laparoscopic adrenalectomy for adrenocortical carcinoma based on completeness of resection, site and timing of initial tumor recurrence, and survival in stage II patients.” Second, distinguishing adenomas from carcinomas could help direct chemotherapy if neoadjuvant treatment is considered prior to definitive surgery. A recent German nonrandomized, single-arm study indicated that neoadjuvant chemotherapy may play a role in certain children with adrenocortical carcinomas, such as those that are unresectable at the time of diagnosis; however, in general, surgical resection of the tumor should be performed as soon as possible after diagnosis [17]. Almost half of our subjects with carcinomas presented with metastatic disease, most often to the liver and lung. These are the two most common sites of metastases described in the existing literature as well [1, 4]. Surprisingly, no relationship between size of the primary tumor and the presence of distant metastatic disease was identified. Although only three of 15 children with carcinoma demonstrated inferior vena cava invasion, all of these children experienced pulmonary metastatic disease as well as local retroperitoneal recurrences. In some instances, tumor thrombus extends into the heart [18]. Based on available follow-up data, the median survival of children with carcinoma was 27 months. At the time of publication 8 of 15 children had died, while 4 were alive and 3 were lost to follow-up after varying lengths of follow-up. In the end, and based in part on prior studies, overall survival in children with adrenocortical neoplasms likely relates to a variety of factors, including patient age, histological features of the lesion, tumor biology/genetics, and tumor size (weight) [6, 19]. Remarkably, all adrenocortical lesions confirmed to be adenomas by histopathology behaved in a benign manner suggesting a correct diagnosis, despite the fact that distinguishing benign from malignant adrenocortical neoplasms in the pediatric population has been historically difficult [1, 2]. This suggests that the ability of pathologists to correctly discriminate adrenocortical adenomas from carcinomas may have substantially improved in recent years based on better diagnostic criteria. All children with adenomas were alive at the time of this report based on available records. Our study has limitations. First, we have a relatively small number of pediatric adrenocortical lesions (n=25), a fact that restricts the conclusions we can make from our data. However, our study is still one of the largest investigations documenting and comparing the imaging features of adenomas and carcinomas in children and adolescents given the rarity of these neoplasms. Second, many more children had undergone CT evaluation than MRI evaluation. Thus, we have insufficient data from which to draw conclusions regarding specific MRI findings that might discriminate adenomas from
carcinomas, such as apparent diffusion coefficient value or loss of signal on T1-weighted gradient recalled echo out-ofphase imaging. Third, we are unable to adequately assess tumor washout of contrast material based on multi-phase CT, a technique that is fundamental to characterizing adrenocortical masses (in particular, adenomas) in adults. Necessary imaging data for washout percentage calculation was available for only two children because pediatric CT examinations performed at our institution are most often a single portal venous phase in order to adhere to “Image Gently” and “as low as reasonably achievable” (ALARA) principles. Finally, in a small number of children (three) with carcinoma we have incomplete follow-up, although Social Security Administration data suggest that these children are still alive.
Conclusion Pediatric adrenocortical carcinomas are larger and more heterogeneous than adenomas and commonly contain calcification. Using the tumor diameter and estimated volume cut-off values mentioned above, CT and MRI discriminate carcinomas from adenomas with a good degree of accuracy (areas under the curve=0.92 and 0.92, respectively); however very small carcinomas and very large adenomas occur, making complete discrimination of these two histological entities impossible based on conventional imaging findings. Further research is needed to determine whether more advanced imaging techniques, such as dynamic contrast-enhanced, chemicalshift and diffusion-weighted MRI, can allow definitive discrimination of adenomas from carcinomas in children and adolescents.
Conflicts of interest None
References 1. Agrons GA, Lonergan GJ, Dickey GE et al (1999) Adrenocortical neoplasms in children: radiologic-pathologic correlation. Radiographics 19:989–1008 2. Wieneke JA, Thompson LD, Heffess CS (2003) Adrenal cortical neoplasms in the pediatric population: a clinicopathologic and immunophenotypic analysis of 83 patients. Am J Surg Pathol 27: 867–881 3. Daneman A, Chan HS, Martin J (1983) Adrenal carcinoma and adenoma in children: a review of 17 patients. Pediatr Radiol 13:11–18 4. Ribeiro J, Ribeiro RC, Fletcher BD (2000) Imaging findings in pediatric adrenocortical carcinoma. Pediatr Radiol 30:45–51 5. Hubertus J, Boxberger N, Redlich A et al (2012) Surgical aspects in the treatment of adrenocortical carcinomas in children: data of the GPOH-MET 97 trial. Klin Padiatr 224:143–147 6. Michalkiewicz E, Sandrini R, Figueiredo B et al (2004) Clinical and outcome characteristics of children with adrenocortical tumors: a
Pediatr Radiol report from the International Pediatric Adrenocortical Tumor Registry. J Clin Oncol 22:838–845 7. Neblett WW, Frexes-Steed M, Scott HW Jr (1987) Experience with adrenocortical neoplasms in childhood. Am Surg 53:117–125 8. Hanna AM, Pham TH, Askegard-Giesmann JR et al (2008) Outcome of adrenocortical tumors in children. J Pediatr Surg 43:843–849 9. Hanson JA, Weber A, Reznek RH et al (1996) Magnetic resonance imaging of adrenocortical adenomas in childhood: correlation with computed tomography and ultrasound. Pediatr Radiol 26:794–799 10. Israel GM, Korobkin M, Wang C et al (2004) Comparison of unenhanced CT and chemical shift MRI in evaluating lipid-rich adrenal adenomas. AJR Am J Roentgenol 83:215–219 11. Caoili EM, Korobkin M, Francis IR et al (2002) Adrenal masses: characterization with combined unenhanced and delayed enhanced CT. Radiology 222:629–633 12. Korobkin M, Brodeur FJ, Francis IR et al (1998) CT time-attenuation washout curves of adrenal adenomas and nonadenomas. AJR Am J Roentgenol 170:747–752 13. Sarwar ZU, Ward VL, Mooney DP et al (2004) Congenital adrenocortical adenoma: case report and review of literature. Pediatr Radiol 34:991–994
14. Miller BS, Doherty GM (2014) Surgical management of adrenocortical tumours. Nat Rev Endocrinol 10:282–292 15. Miller BS, Gauger PG, Hammer GD et al (2012) Resection of adrenocortical carcinoma is less complete and local recurrence occurs sooner and more often after laparoscopic adrenalectomy than after open adrenalectomy. Surgery 152:1150–1157 16. Else T, Williams AR, Sabolch A et al (2014) Adjuvant therapies and patient and tumor characteristics associated with survival of adult patients with adrenocortical carcinoma. J Clin Endocrinol Metab 99:455–461 17. Redlich A, Boxberger N, Strugala D et al (2012) Systemic treatment of adrenocortical carcinoma in children: data from the German GPOH-MET 97 trial. Klin Padiatr 224:366–371 18. Godine LB, Berdon WE, Brasch RC et al (1990) Adrenocortical carcinoma with extension into inferior vena cava and right atrium: report of 3 cases in children. Pediatr Radiol 20:166–168, discussion 169 19. Bugg MF, Ribeiro RC, Roberson PK et al (1994) Correlation of pathologic features with clinical outcome in pediatric adrenocortical neoplasia. A study of a Brazilian population. Brazilian Group for Treatment of Childhood Adrenocortical Tumors. Am J Clin Pathol 101:625–629
