RESEARCH
Check for updates
Pediatric Adrenocortical Neoplasms: A Study Comparing Three Histopathological Scoring Systems
Hemlata Jangir1 . Isheeta Ahuja1 . Shipra Agarwal1 (D . Vishesh Jain2D . Jagdish Prasad Meena3D. Sandeep Agarwala2 . Rajni Sharma4 . Mehar Chand Sharma1(D . Venkateswaran K. lyer1D . Kalaivani Mani5
Accepted: 14 April 2023 / Published online: 9 May 2023 @ The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2023
Abstract
Adrenocortical neoplasms are rare in childhood. Their histopathological categorization into benign and malignant is often challenging, impacting further management. While the AFIP/Wieneke scoring system is widely used for the prognostic clas- sification of these tumors, it has limitations. Few other tumor scoring systems have evolved over the past few years. These have been validated in adults but not yet in pediatric patients. We evaluated a cohort of pediatric adrenocortical neoplasms to assess the applicability of AFIP/Wieneke criteria and the recently introduced Helsinki score and reticulin algorithm in predicting clinical outcomes. A tumor was considered ‘clinically aggressive’ in the presence of any of the following: metas- tases, recurrence, progressive disease, or death due to disease. Cases without any such event were considered ‘clinically good’. Event-free survival time was the duration from the date of clinical presentation to any post-operative adverse event. For overall survival analysis, the endpoint was either the last follow-up or death due to disease.Using ROC curve analysis, the obtained cut-off Helsinki score of 24 could stratify the cases into two prognostically relevant groups. Survival analysis showed significant differences in the event-free and overall survival of these two groups of patients, validating the proposed cut-off. None of the three histopathological scoring systems could predict an unfavorable outcome with 100% accuracy. All showed a sensitivity of ≥80%, with the reticulin algorithm achieving 100% sensitivity. The specificity and accuracy of the AFIP/Wieneke criteria were the lowest (62.5% and 73.08%, respectively). While the Helsinki score (at the cut-off score of 24) and the reticulin algorithm had similar accuracy rates (80.77%, and 80%, respectively), the specificity of the former was higher (81.25%) than the latter (68.75%). A separate analysis revealed that the Ki-67 index at a cut-off of 18% had a sensitivity of 80% and a specificity of 81.25% for predicting an unfavorable outcome.
Keywords Adrenocortical neoplasm · Adrenocortical carcinoma · Pediatric · AFIP/Wieneke criteria · Helsinki score . Reticulin Algorithm · Linn-Weiss-Bisceglia criteria · Ki-67 · PD-L1
| ☒ Shipra Agarwal drshipra0902@gmail.com | Venkateswaran K. Iyer iyer_venkat@hotmail.com Kalaivani Mani manikalaivani@gmail.com | |
|---|---|---|
| Hemlata Jangir dr.hem2507@yahoo.com | ||
| Isheeta Ahuja isheetaahuja@gmail.com | 1 | Department of Pathology, All India Institute of Medical Sciences (AIIMS), New Delhi 110029, India |
| Vishesh Jain dr.vishesh79@gmail.com | 2 | Department of Pediatric Surgery, All India Institute of Medical Sciences, New Delhi, India |
| Jagdish Prasad Meena drjpmeena@gmail.com | 3 | Department of Pediatric Medical Oncology, All India Institute of Medical Sciences, New Delhi, India |
| Sandeep Agarwala sandpagr@hotmail.com | 4 | Department of Pediatric Endocrinology, All India Institute of Medical Sciences, New Delhi, India |
| Rajni Sharma drrajnisharma@yahoo.com | 5 | Department of Biostatistics, All India Institute of Medical Sciences, New Delhi, India |
| Mehar Chand Sharma sharmamehar@yahoo.co.in |
Introduction
Adrenocortical carcinomas (ACC) are rare in the pediatric age group. Their incidence is 1.5 million cases/year, con- stituting 0.2% of all pediatric malignancies in the West [1]. A similar proportion of 0.1% has been recorded in India, accounting for 4% of all pediatric malignancies [2].
The Weiss criteria, in their original and the modified ver- sions, are most commonly used for the separation of adult- onset adrenocortical neoplasms (ACN) into adrenocortical adenoma (ACA) and ACC [3-6]. Two new scoring systems, the Helsinki system [7] and the reticulin algorithm, have been proposed for the risk stratification and classification of ACN [8]. Pennanen et al., proposed that Helsinki scores of 8.5 and 17 can, respectively, predict metastases and poor survival in adult ACC patients [7]. However, a follow-up validation study found that Helsinki scores of 13 and 19 could divide the cases into three distinct prognostic groups [9]. Another classification system used for oncocytic ACN is the Linn-Weiss-Bisceglia criteria (LWB) [10]. The Hel- sinki system is apparently applicable to oncocytic tumors too [9]. However, reports have suggested a limited utility of the reticulin algorithm in these tumors [11-13].
As the Weiss criteria lead to overdiagnosis of malig- nancy in pediatric cases [7, 14], Wieneke et al. from the Armed Forces Institute of Pathology (AFIP) provided a scoring system that is currently the most accepted system for these patients [15-21]. A recent study on ACC involv- ing 57 patients aged < 18 years documented the classifica- tion system to have 77% accuracy [22].
The Helsinki score has also been assessed for its util- ity in the prognostication of pediatric ACC, but only to a limited extent and as a part of a cohort containing mainly adult patients [9]. The reticulin algorithm has, however, not yet been evaluated in this age group [8, 11, 12, 23-25]. There is also the dilemma of which classification system to be used in pediatric oncocytic ACN. Most of the avail- able literature is based on a handful of case reports or short series or as part of a cohort of adult tumors [26-29], where these have been classified using the LWB score.
In the current study, the applicability of the Helsinki score and the reticulin algorithm was assessed in a cohort of pediatric ACN cases, and the results were compared with the AFIP/Wieneke system. A cut-off Helsinki score for risk-stratifying pediatric patients was also calculated.
As a preliminary work-up for predicting response to immunotherapy in childhood ACC, the tumors were also evaluated for the immunohistochemical expression of pro- grammed cell death ligand-1 (PD-L1), a novel target for cancer immunotherapy [30].
Materials and Methods
The retrospective single center study was performed in the department of Pathology, All India Institute of Medical Sci- ences, New Delhi in collaboration with the departments of pediatric surgery, pediatric oncology, and pediatric endocri- nology after obtaining appropriate ethical clearance from the institute ethics committee (IEC-536/06.08.2021).
A total of 28 cases with a diagnosis of ACN, involving patients aged ≤ 18 years, were retrieved from the departmen- tal archives (2015-2022). Two cases lacking representa- tive tumor blocks or clinical details were excluded. Of the remaining 26 patients, there were 25 resection specimens and one core needle biopsy from an unresectable tumor.
The tumor slides were reviewed by three pathologists (MCS, VI, SA), two with interest in Endocrine Pathology (SA, MCS), and one in Pediatric Pathology (VI). The tumors were subtyped into conventional, mixed (50-90% oncocytic cell population), and ‘pure’ oncocytic ACN (> 90% oncocytic cells) [31]. Details including the weight and dimensions of the tumor were retrieved from the archived histopathology requisition forms. All the tumors were scored and classi- fied following the AFIP/Wieneke criteria [6, 15], Helsinki score [6, 7], and reticulin algorithm [6, 8], irrespective of the extent of oncocytic morphology. The ‘pure’ oncocytic neoplasms were also classified using the LWB criteria [6, 10]. Mitoses were counted in hotspot areas [6].
Immunohistochemistry
Four micrometer thick sections were cut from representative tumor blocks. Reticulin staining was done using the Gordon- Sweet silver histochemistry. Immunohistochemistry (IHC) for Ki-67 (clone SP6, Invitrogen) was performed manually. PD-L1 IHC (clone SP263, VENTANA Medical Systems, Inc) was done on the fully automated VENTANA Bench- mark XT platform, optimized with the OptiView DAB IHC Detection kit (VENTANA Medical Systems, Inc).
The reticulin staining was interpreted as per the already established criteria in adults [8, 24, 25]. Significant loss of the reticulin network was defined as loss of continuity of reticular fibers over at least 25% of the evaluated tumor area [24, 25]. A minimum of 500 tumor cells were evaluated for the IHC stains. The Ki-67 labeling index was calculated as percentage positivity in hotspots. PD-L1-positivity was expressed as tumor proportion score (TPS), calculated as the percentage of tumor cells showing membranous positivity (partial or complete, and of any intensity). A TPS of 1% was used as a cut-off to consider a case as PD-L1- positive.
Tumor Scoring Systems
Table 1 details the parameters assessed and criteria used for assessing the tumors using the four histopathological scor- ing systems.
Clinical Follow-Up
The age, sex, serum hormone profile, management, and follow-up of all the patients were obtained from the medi- cal records.
The clinical outcome was considered the gold standard. A tumor was considered ‘clinically aggressive’ if there were any of the following events: metastases, recurrence (reap- pearance of disease after a complete response, in the same or a different anatomic site), progressive disease (increase by ≥ 20% of at least one measurable lesion or appearance of a new lesion), or death due to disease. Rest of the cases were considered ‘clinically good’. Event-free survival (EFS) time was calculated as the duration from the date of clinical presentation to any adverse event at the time of
post-operative follow-up. Similarly, the endpoint for overall survival (OS) analysis was either the last follow-up or death due to disease.
Statistical Analysis
Statistical analysis was performed on STATA 16.0. Cut-off Helsinki and Ki-67 scores to predict the clinical outcome were calculated using logistic regression and receiver operating characteristic (ROC) analyses. The area under the ROC curve (AUC) was used to summarise the discrimination power of the test. The cut-off score stratified the patients into two prog- nostically different groups. The event-free and overall survival characteristics of these two groups were analyzed by plotting the Kaplan-Meier survival curves. A p-value less than 0.05 was considered statistically significant. The hazard ratios at different Helsinki and Ki-67 scores were also calculated.
To determine the best scoring system applicable to pedi- atric ACN, the sensitivity, specificity, positive likelihood ratio, and accuracy of the three systems for predicting clini- cal outcome were also assessed.
| Histopathological scoring system | Parameters | Score | Interpretation |
|---|---|---|---|
| AFIP/Wieneke criteria [6, 15] | Tumor weight>400 gm | 1 | Score ≤2: Benign |
| Tumor size > 105 mm | 1 | Score 3: Uncertain malignant potential | |
| Extension into peri-adrenal soft tissue | 1 | Score ≥4: Malignant/poor outcome | |
| Inferior vena cava invasion | 1 | ||
| Venous invasion | 1 | ||
| Capsular invasion | 1 | ||
| Tumor necrosis | 1 | ||
| Mitotic count> 15/4mm2 field area | 1 | ||
| Atypical mitoses | 1 | ||
| Helsinki score [6, 7] | Mitotic rate > 5/10 mm2 | 3 | Score=Ki-67 (%)+3xMitosis +5xNecrosis |
| Necrosis | 5 | ||
| Ki-67 labeling index | Numeric value | ||
| from hotspot | |||
| Reticulin algorithm [6, 8] | Altered reticulin framework (≥25% of malignancy: | tumor area) in | association with one of the following features indicates |
| Mitotic count> 5/10 mm2 | |||
| Tumor necrosis | |||
| Vascular invasion | |||
| Linn-Weiss-Bisceglia scoring system [6, 10] | Major criteria: | Absence of any major and minor criteria: Benign | |
| Mitoses>5/10mm2 | Presence of any minor criterion: Borderline Presence of any major criterion: Malignant | ||
| Atypical mitoses | |||
| Venous invasion | |||
| Minor criteria | |||
| Tumor size > 100 mm | |||
| Tumor weight> 200 gm | |||
| Tumor necrosis | |||
| Capsular invasion | |||
| Sinusoidal invasion |
Results
Clinical Details
The median age of the patients at the time of presentation was fifty-five months (range; 6-204 months). There was a slight male preponderance (M:F=1.3:1).
Eighteen of the 26 (69.2%) tumors were secretory (Supplementary table). Five patients had tumor spillage at the time of surgery. On follow-up, 10 (38.5%) behaved aggressively. Chemotherapy was given to seven patients. At the end of follow-up, 15 (57.7%) patients were alive without disease, one (3.8%) was alive with disease (AWD), and seven patients (26.9%) died of disease (DOD). Three patients were lost to follow-up after an initial follow-up period of 12, 14, and 47 months, respectively. The median EFS and OS was 16.4 months (range, 1.8-97.1 months) and 26.4 months (range, 3.5-97.1 months), respectively.
Pathological Analysis
The mean size and weight of the tumors were 7.1 cm (range; 1-17.5 cm), and 110.3 gm (range; 7.5-750 gm). On histo- pathological evaluation, there were 13 (50%) conventional ACN, nine (34.6%) ‘pure’ oncocytic ACN, and four (15.4%) mixed ACN (Fig. 1).
Tumor Scoring System
Using the AFIP/Wieneke criteria, 11 tumors (42.3%) were classified as benign, and 15 (57.7%) as malignant. The Hel- sinki scores varied from 0.5 to 78 (median; 14). Reticulin
stain was performed on all except the needle biopsy. Fol- lowing the reticulin algorithm, 11 (44%) were categorized benign and 14 (56%) malignant.
LWB criteria classified one (11.1%) of the nine oncocytic ACNs benign and the remaining (88.9%) malignant.
The clinical and pathological details of individual cases including the results of the tumor scoring systems have been provided in Supplementary table.
Cut-Off Values of the Helsinki Score
Using ROC curve analysis, a Helsinki score of 24 predicted the clinical outcome with 80% (95% CI=49.02-94.33) sen- sitivity and 81.25% (95% CI=56.99-93.41) specificity (AUC 0.93; 95% CI=0.84-1.00) (Table 2, Fig. 2). The hazard ratios at this cut-off for unfavorable outcome and disease-related death were 10.28 (95% CI=2.08-50.71; p=0.004) and 12.07 (95% CI=2.35-61.96; p=0.003), respectively (Table 3).
Table 3 details the sensitivity, specificity and hazard ratios of the Helsinki score and the Ki-67 index at different values.
Cut-Off Values of the Ki-67 Labeling Index
Similarly, a Ki-67 index of 18% was helpful in prognostica- tion with the same sensitivity, specificity, and hazard ratio as the Helsinki score of 24 (AUC 0.91; 95% CI=0.79-1.00) (Table 2, Fig. 2).
Survival Analysis
Using the Helsinki score of 24, the tumors were re-grouped into two prognostic groups. Fifteen (57.7%) cases had a
A
B
C
D
E
F
| True Positive | False Positive | True Negative | False Negative | Sensitivity (95% CI) | Specificity (95% CI) | Accuracy (95% CI) | Positive likelihood ratio | Area under the curve (95% CI) | |
|---|---|---|---|---|---|---|---|---|---|
| AFIP/ | 9 (34.6%) | 6 (23.1%) | 10 (38.5%) | 1 (3.8%) | 90% | 62.5% | 73.08% | 2.4 | - |
| Wieneke | (59.58- | (38.64- | (53.92- | (1.69- 3.41) | |||||
| scoring system | 98.21) | 81.52) | 86.3) | ||||||
| Helsinki | 8 (30.8%) | 3 | 13 (50%) | 2 (7.7%) | 80% | 81.25% | 80.77% | 4.27 (2.09 | 0.93 (0.84- |
| score | (11.5%) | (49.02- | (56.99- | (62.12- | -8.72) | 1.00) | |||
| (using score 24) | 94.33) | 93.41) | 91.49) | ||||||
| Reticulin | 9 (36%) | 5 (20%) | 11 (44%) | 0 | 100% | 68.75% | 80% (60.87- | 3.2 | - |
| Algorithm | (70.08-100) | (44.4- | 91.14) | (2.16-4.74) | |||||
| 85.84) | |||||||||
| Ki-67 index | 8 (30.8%) | 3 (11.5%) | 13 (50%) | 2 (7.7%) | 80% | 81.25% | 80.77% | 4.27 (2.09 | 0.91 (0.79- |
| (using | (49.02- | (56.99- | (62.12- | -8.72) | 1.00) | ||||
| score 18) | 94.33) | 93.41) | 91.49) |
Helsinki score of <24, and 11 (42.3%) had a ≥ 24 score (Fig. 3). Survival analysis revealed significant differences in the EFS (p=0.0006) and the OS (p=0.0003) between the two groups (Fig. 4).
Sensitivity, Specificity and Diagnostic Accuracy of the Scoring Systems
The AFIP/Wieneke criteria, the Helsinki score of 24, and the reticulin algorithm had a sensitivity ranging from 80 to 100% and a specificity from 62.5 to 81.25% (Table 2).
Oncocytic Adrenocortical Neoplasms
The small number (n=9) of oncocytic tumors in the cohort limits the reliability of the statistical analysis in this sub- group. Of these nine patients, seven did not develop any adverse event, and two had an unfavorable outcome. Using the LWB criteria, three of these tumors (one ‘clinically good’, and two ‘clinically aggressive’) were correctly clas- sified (Supplementary table, Fig. 3). The AFIP/Wieneke criteria and the reticulin algorithm appropriately character- ized five and seven cases, respectively. The Helsinki scores
Fig. 2 ROC curves of the Ki-67 labeling index and the Helsinki score
1.00
0.75
Sensitivity
0.50
0.25
0.00
0.00
0.25
0.50
0.75
1.00
1-specificity
Ki-67: AUC=0.9094
Helsinki score: AUC=0.9312
Reference
Table 3 Sensitivity, Specificity and Hazard ratios of the Helsinki score and the Ki-67 index at different values
Helsinki score
| Cut off | Sensitivity (95% CI) | Specificity (95% CI) | Hazard Ratio (95% CI) for EFS | P value for EFS | Hazard Ratio (95% CI) for OS | P value for OS |
|---|---|---|---|---|---|---|
| ≥15 | 90% | 75% | 18.50 | 0.008 | 10.36 | 0.004 |
| (59.58-98.21) | (50.5-89.82) | (2.18-157.18) | (2.11-50.78) | |||
| ≥23 | 80% | 75% | 9.16 | 0.007 | 11.94 | 0.003 |
| (49.02-94.33) | (50.5-89.82) | (1.83-45.76) | (2.32-61.48) | |||
| ≥24 | 80% | 81.25% | 10.28 | 0.004 | 12.07 (2.35-61.96) | 0.003 |
| (49.02-94.33) | (56.99-93.41) | (2.08-50.71) |
| Cut off | Sensitivity (95% CI) | Specificity (95% CI) | Hazard Ratio (95% CI) for EFS | P value for EFS | Hazard Ratio (95% CI) for OS | P value for OS |
|---|---|---|---|---|---|---|
| ≥ 15% | 80% | 75% | 9.16 | 0.007 | 11.94 | 0.003 |
| (49.02-94.33) | (50.5-89.82) | (1.83-45.76) | (2.32-61.48) | |||
| ≥ 18% | 80% | 81.25% | 10.28 | 0.004 | 12.07 | 0.003 |
| (49.02-94.33) | (56.99-93.41) | (2.08-50.71) | (2.35-61.96) | |||
| ≥20% | 70% | 87.5% | 8.64 | 0.002 | 7.44 | 0.006 |
| (39.68-89.22) | (63.98-96.5) | (2.17-34.38) | (1.78-31.15) |
of the seven ‘clinically good’ tumors were 2, 3, 7, 9, 24, 26, and 28, and of the aggressive ones 15, and 26, respectively (Supplementary table and Fig. 3). Using the generated cut- off value of 24, the Helsinki system could, hence, correctly classify five of the nine cases.
PD-L1 Immunostaining Analysis
PD-L1 immunostaining was performed in 25 of the 26 tumors. All the cases were PD-L1 negative. TPS was <1% in one case and zero in the rest.
Helsinki score
0.5
1
2
3
3
3
5
7
9
10
11
11
13
15
23
24
26
26
28
35
45
50
54
61
68
78
Ki-67 labeling index
0.5
1
2
3
3
3
2
7
1
5
6
8
13
7
15
21
18
18
20
27
37
42
46
53
60
70
AFIP/Wieneke criteria
Reticulin algorithm
ND
LWB criteria
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
Metastasis
Recurrence
Progressive disease
Tumor spillage
Chemotherapy
Follow-up
Tumor behavior
Tumor subtype
Patient Age(months)
54
84
156
12
56
120
24
24
12
54
204
57
35
60
6
24
30
30
7
120
72
144
166
48
61
84
Patient ID
6
5
8
7
11
17
12
19
3
10
14
15
18
2
23
22
1
21
13
16 9
25
24
26
4
20
Helsinki score ☐ -24
AFIP/Wieneke criteria
Reticulin algorithm
LWB
score
Metastasis
Recurrence
Progressive disease
Tumor spillage
Chemotherapy
Follow-up
Tumor behavior
Tumor subtype
☐ Benign
☐ Benign
☐ Good
☐ ≥24
☐ Benign
☐ Absent
☐ Absent
☐ Given
☐ Malignant
☐ Absent
☐ Present
☐ Present
☐ Absent
☐ Malignant
☐ Malignant
☐ Present
☐ Present
☐ Not given
☐ Free of disease
☐ Conventional
☐ Alive with disease
☐ Aggressive
☐ Oncocytic
ND ☒ Not done
NA ☒
Not applicable
☐ Lost
☐ Death
☐ Mixed
☒ Springer
A
1.00
Kaplan-Meier survival estimates
B
1.00
Kaplan-Meier survival estimates
Event free survival
0.75
Prob. of survival
0.75
0.50
Pvalue = 0.0006
0.50
Pvalue = 0.0003
0.25
0.25
0.00
0.00
0
12
24
36
60
0
12
24
36
48
60
72
84
96
Time in months
48
72
84
96
Time in months
Number at risk
Number at risk Helsinki score<24 15
Helsinki score<24
15
12
11
9
6
5
3
1
1
14
12
10
6
5
3
1
1
-Helsinki score>=24
11
4
1
0
0
0
0
0
0
Helsinki score>=24 11
8
2
1
0
0
0
0
0
Discussion
In ACNs, a histopathological classification system having an absolute agreement with clinical outcome remains elu- sive. As no single histopathological characteristic perfectly correlates with the clinical outcome and disease prognosis, most currently-used classification systems are based on mul- tiple clinicopathological parameters. The problem is worse in pediatric cases.
The novel Helsinki scoring system, validated by Pennanen et al., on 177 adult ACC patients, has been proposed to be predictive of metastases at a cut-off of 8.5 with 100% sensi- tivity and 99.4% specificity. Cases with Helsinki score> 8.5 behaved in a malignant way. Further, patients with scores> 17 had a poorer OS [7]. Subsequently, Duregon et al., evaluated the utility of the Helsinki score on a larger series of 225 ACC. They reported that Helsinki scores of 13 and 19 could re- classify ACC into three prognostic groups having significantly different OS (p<0.0001). Importantly, besides adult-onset tumors with conventional morphology, they also included oncocytic and myxoid ACCs, as well as a subset of pediat- ric cases. Another cut-off of> 28.5 was suggested to predict death due to disease [9]. Renaudin evaluated the prognostic relevance of these cut-off scores in 43 adult cases of oncocytic ACN. Five of the 13 tumors with a score of> 8.5 had a score of ≥ 13 and had a poor clinical outcome. The only two patients who died of disease had a Helsinki score ≥ 19. Hence, in their study, the cut-off values of 13 and 19 performed better than 8.5 and 17 [23]. In another recent study on 37 adult patients, ACCs had a significantly higher median Helsinki score (28; range: 10-56) than ACA (3; range: 1-5). Although the Weiss and the Helsinki scores had a significant positive correla- tion, none had any statistically significant association with progression-free survival or OS [25]. The Reticulin algorithm, proposed by Volante et al., is another classification system.
In their study, the algorithm was 100% sensitive and specific in identifying all cases with Weiss score ≥ 3 as malignant [8]. The algorithm was subsequently validated in two different adult patient-based studies [24, 25].
To date, there are no pediatric ACN-centric studies evaluating the utility of the Helsinki scoring system and the reticulin algorithm in the classification and prognos- tication of these tumors. In the current study, using ROC curve analysis, a cut-off Helsinki score of 24 was best pre- dictive of an aggressive clinical outcome. All but two tumors (86.7%; 13/15) with a Helsinki score of <24 were ‘clini- cally good’. Eight of the 11 (72.7%) with a score ≥ 24 had a poor clinical outcome. The median EFS of the clinically good and the aggressive cases were 37.8 months (range, 8.1-97.1 months) and 6.2 months (range, 1.8-31.1 months), respectively. Similarly, their OS duration was 37.8 months (range; 8.1-97.1 months) and 16.7 months (range; 3.5-47.3 months), respectively. The hazard of developing an adverse event in tumors with a Helsinki score of 24 and above was 10.3 times that of those with lower scores. Simi- larly, the former group of patients was about twelve times more likely to die of the disease.
All three multi-parameter scoring systems, the AFIP/Wie- neke criteria, Helsinki score, and the reticulin algorithm, performed well in the histologic classification of pediatric ACN. The sensitivity of all three was > 80%, with the reti- culin algorithm achieving 100% sensitivity. The specificity and accuracy were highest for the Helsinki score (81.25% and 80.77%), followed by the reticulin algorithm (68.75% and 80%), and AFIP/Wieneke criteria (62.5% and 73.08%) (Table 2). Jehangir et al., in their analysis on 22 pediatric patients, found the AFIP/Wieneke system to have a sensi- tivity of 100%, a specificity of 87%, a positive predictive value of 80%, and a negative predictive value of 100% [20]. Similar to our results, a recent systematic review of a larger
cohort of 128 patients reported the AFIP/Wieneke system to have 92% sensitivity and 77% specificity for prognos- tic stratification of these tumors [14]. Other studies have found its accuracy to vary from 77 to 100% [15-22]. Picard analyzed 95 patients of pediatric ACN for the prognostic relevance of the AFIP/Wieneke scoring system. The authors reported that while the score was perfectly predictive of the outcome when the score was <3 (‘benign’), a signifi- cant proportion (43.5%) of their patients characterized as malignant (score>3) had a benign outcome. Hence, the authors concluded that the AFIP/Wieneke system may not be sufficient enough to predict clinical outcome and guide perioperative therapeutic management [32]. In the current study, using AFIP/Wieneke criteria, seven cases were mis- classified, one (4%) as benign and six (23.1%) as malig- nant. The Helsinki score and the reticulin algorithm mis- characterized five cases each (Table 2, Fig. 3). Cases 14 and 2 had a Helsinki score <24 (11 and 15, respectively) but had an unfavorable outcome. The reticulin algorithm correctly classified both the tumors as malignant. Follow- ing the AFIP/Wieneke criteria, while case 14 was ‘malig- nant’, case 2 was ‘benign’. The former patient (case# 14) had been operated on at another center and was referred to us for further management. Only one tumor block was avail- able for histopathological evaluation. Microscopy revealed vascular invasion, necrosis, and atypical mitosis. There was radiological evidence of inferior vena cava invasion, but the mitotic rate (3/4mm2 and 5/10mm2) and Ki-67 index (6%) were relatively low. It is possible that evaluation of more sections for areas with higher proliferative activity would have increased the Helsinki score. However, the reason for the discrepancy remains unexplained in case# 2. There was focal tumor necrosis and occasional atypical mitosis but cap- sular and vascular invasion were absent. While the Ki-67 index was 7%, the mitotic rate was < 15/4mm2 (criterion used in AFIP/Wieneke criteria) and > 5 per 10mm2 (criterion used in the Helsinki scoring system and the reticulin algo- rithm). In pediatric ACC, molecular alterations, like TP53 and CTNNB1 mutations are associated with a worse clinical outcome [33]. Moreover, other factors like age and treatment variables also influence the disease course. Assessment of these is, however, beyond the scope of the current study.
Our cohort included cases with oncocytic histomorphol- ogy. While pediatric ACN are themselves rare, an oncocytic subtype is even more uncommon. Although statistical analy- sis was not feasible, the reticulin algorithm performed the best (7/9 correctly classified), followed by the Helsinki score (5/9 correctly classified) and the AFIP/Wieneke criteria (5/9 correctly classified). The LWB, in contrast, correctly cat- egorized only three of these cases. Duregon et al., suggested that the reticulin alteration may be focal in oncocytic ACN
requiring careful evaluation and staining of more than one tumor section [11]. In their subsequent study on 245 ACN, including 44 oncocytic tumors, the reticulin stain worked well, irrespective of the histological subtype [12]. However, in a recent series on adult patients, the reticulin algorithm classified 56% of the benign oncocytic ACN, including both pure and mixed forms, as malignant. All three cases with poor outcome were categorized as malignant. Further, it was observed that atypical mitoses was the sole fulfilled major criterion in three of their six cases that showed discordance between the reticulin algorithm and the other scoring sys- tems (malignant with LWB score vs. benign with reticulin algorithm and Helsinki score). The authors concluded that the Helsinki score was the best for classifying oncocytic tumors [23]. In our study, atypical mitoses was the only major LWB criterion present in two of the six cases misdi- agnosed as malignant. Notably, four of the nine ‘clinically good’ tumors having conventional histomorphology also showed atypical mitotic forms.
The Ki-67 labeling index of a tumor has a major impact on the final Helsinki score. In the current study, using ROC curve analysis, a Ki-67 cut-off of 18% achieved a sensitivity and specificity of about 80%. Martins-Filho, in their cohort of 44 pediatric tumors, found that a Ki-67 index ≥ 15% predicted recurrence and/or poor outcome with a sensitivity of 89% and a specificity of 77%. When < 10%, the index ruled out malignancy, at a sensitivity of 100% and a specificity of 51% [34]. At 15% cut-off, we obtained a sensitivity of 80% and a specificity of 75% (Table 3). In the study by Picard et al., none of the cases with a Ki-67 LI <15% had disease recur- rence [32]. However, we found two cases with Ki-67 index of 6% (case 14) and 7% (case 2) who developed adverse events (Fig. 3). Hence, despite a similar cut-off, the lower limit of Ki-67 index was different. This may be explained by dif- ferences in the antibody clone, immunostaining technique, method of evaluation and/or cohort of cases.
Immunotherapy targeting the PD-1/PD-L1 axis is increas- ingly being used for the management of many recalcitrant malignancies. Immunoexpression of PD-L1 protein on the sur- face of the tumor cells helps in predicting the tumor response to immunotherapeutic agents. In the case of ACC, only a handful of studies have evaluated PD-L1 immunoexpression, mainly in adults [35-37], with the proportion of cases positive for PD-L1 ranging from none [36] to 21% [37]. In the latter study, the authors also documented a limited response to Pembrolizumab irrespective of the status of tumor PD-L1 expression [37]. In our study, none of the cases was PD-L1-positive. Interestingly, Geoerger reported partial response to Pembrolizumab in two of the four paediatric patients with ACC. Hence, larger-scale studies and clinical trials are essential for confirming the effec- tivity of immunotherapy in these patients [38].
Conclusions
This study analyzed the utility of the widely-used AFIP/Wie- neke scoring system, the newly postulated Helsinki score, and the reticulin algorithm in predicting an unfavorable outcome in pediatric adrenocortical neoplasms. We propose and validate a cut-off Helsinki score applicable to these patients. A score of≥24 could classify these tumors into two statistically rele- vant prognostic groups, that showed significant differences in the event-free and overall survival rates. Although none of the three scoring systems could predict an unfavorable outcome with 100% accuracy, the Helsinki score (at the cut-off of 24) and the reticulin algorithm achieved an accuracy of ≥ 80%. Importantly, the reticulin algorithm showed 100% sensitivity. Highest specificity was seen with the Helsinki scoring system. Notably, the current study could also demonstrate the applicabil- ity of the reticulin algorithm in pediatric oncocytic tumors, where it performed the best in comparison to the other scoring systems.
The Ki-67 proliferative index, in itself, is also a robust parameter. At a cut-off of ≥ 18%, it achieved ≥80% sensitiv- ity and specificity, but it may be better to combine the index with other parameters as in the Helsinki scoring system.
Our study is limited by a relatively small case number and a lack of molecular profiling. Moreover, it is not possible to rule out if any of the tumors classified as ‘clinically good’ would develop a recurrence on further follow-up. Hence, multi-institutional studies with longer follow-up would help confirm our findings.
Supplementary Information The online version contains supplemen- tary material available at https://doi.org/10.1007/s12022-023-09767-z.
Acknowledgements We thank the technical staff, especially Mr. Rahul S.R., and Mrs. Bimla Devi, of the Department of Pathology, AIIMS, New Delhi for their cooperation in sectioning and staining including histochemistry for reticulin stain.
Author Contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Hemlata Jangir, Isheeta Ahuja, Jagdish Prashad Meena, Vishesh Jain, Rajni Sharma, Sandeep Agarwala, Kalaivani Mani, and Shipra Agarwal. The first draft of the manuscript was written by Hemlata Jangir, and all authors commented on previous versions of the manu- script. All authors read and approved the final manuscript.
Availability of Data and Material The data used in the manuscript has been made available as supplementary files.
Declarations
Ethical Approval The study was approved by the institute ethics com- mittee (All India Institute of Medical Sciences, New Delhi, India) (IEC-536/06.08.2021).
Informed Consent Not applicable.
Conflict of Interest The authors declare no competing interests direct- ly or indirectly related to the work submitted for publication.
References
1. Golden SH, Robinson KA, Saldanha I, Anton B, Ladenson PW (2009) Clinical review: prevalence and incidence of endocrine and metabolic disorders in the United States - a comprehensive review. J Clin Endocrinol Metab 94: 1853-1878.
2. National Cancer Registry Programme (ICMR-NCDIR), https:// ncdirindia.org/All_Reports/Report_2020/Factsheet/Fact_Sheet_ 2020.pdf.
3. Weiss LM (1984) Comparative histologic study of 43 metas- tasizing and nonmetastasizing adrenocortical tumors. Am J Surg Pathol 8:163-169. https://doi.org/10.1097/00000478- 198403000-00001.
4. Weiss LM, Medeiros LJ, Vickery AL Jr (1989) Pathologic features of prognostic significance in adrenocortical carci- noma. Am J Surg Pathol 13:202-206. https://doi.org/10.1097/ 00000478-198903000-00004.
5. Aubert S, Wacrenier A, Leroy X, Devos P, Carnaille B, Proye C, Wemeau JL, Lecomte-Houcke M, Leteurtre E (2002) Weiss sys- tem revisited: a clinicopathologic and immunohistochemical study of 49 adrenocortical tumors. Am J Surg Pathol 26:1612-1619. https://doi.org/10.1097/00000478-200212000-00009.
6. Mete O, Erickson LA, Juhlin CC, de Krijger RR, Sasano H, Volante M, Papotti MG (2022) Overview of the 2022 WHO Classification of Adrenal Cortical Tumors. Endocr Pathol 33:155-196. https://doi. org/10.1007/s12022-022-09710-8.
7. Pennanen M, Heiskanen I, Sane T, Remes S, Mustonen H, Haglund C, Arola J (2015) Helsinki score-a novel model for pre- diction of metastases in adrenocortical carcinomas. Hum Pathol 46:404-410. https://doi.org/10.1016/j.humpath.2014.11.015.
8. Volante M, Bollito E, Sperone P, Tavaglione V, Daffara F, Porpiglia F, Terzolo M, Berruti A, Papotti M (2009) Clinico- pathological study of a series of 92 adrenocortical carcinomas: from a proposal of simplified diagnostic algorithm to prognos- tic stratification. Histopathology 55:535-543. https://doi.org/ 10.1111/j.1365-2559.2009.03423.x.
9. Duregon E, Cappellesso R, Maffeis V, Zaggia B, Ventura L, Berruti A, Terzolo M, Fassina A, Volante M, Papotti M (2017) Valida- tion of the prognostic role of the “Helsinki Score” in 225 cases of adrenocortical carcinoma. Hum Pathol 62:1-7. https://doi.org/10. 1016/j.humpath.2016.09.035.
10. Bisceglia M, Ludovico O, Di Mattia A, Ben-Dor D, Sandbank J, Pasquinelli G, Lau SK, Weiss LM (2004) Adrenocortical oncocytic tumors: report of 10 cases and review of the literature. Int J Surg Pathol 12:231-243. https://doi.org/10.1177/106689690401200304.
11. Duregon E, Volante M, Cappia S, Cuccurullo A, Bisceglia M, Wong DD, Spagnolo DV, Szpak-Ulczok S, Bollito E, Daffara F, Berruti A, Terzolo M, Papotti M (2011) Oncocytic adrenocorti- cal tumors: diagnostic algorithm and mitochondrial DNA profile in 27 cases. Am J Surg Pathol 35:1882-1893. https://doi.org/10. 1097/PAS.0b013e31822da401.
12. Duregon E, Fassina A, Volante M, Nesi G, Santi R, Gatti G, Cappellesso R, DalinoCiaramella P, Ventura L, Gambacorta M, Dei Tos AP, Loli P, Mannelli M, Mantero F, Berruti A, Terzolo M, Papotti M (2013) The reticulin algorithm for adrenocortical tumor diagnosis: a multicentric validation study on 245 unpub- lished cases. Am J Surg Pathol 37:1433-1440. https://doi.org/10. 1097/PAS.0b013e31828d387b.
13. Fonseca D, Murthy SS, Tagore KR, Rao BV, Thamminedi SR, Raju KVVN, Sharma R, Challa S (2017) Diagnosis of Adrenocor- tical Tumors by Reticulin Algorithm. Indian J Endocrinol Metab 21:734-737. https://doi.org/10.4103/ijem.IJEM_573_16.
14. Riedmeier M, Thompson LDR, Molina CAF, Decarolis B, Härtel C, Schlegel PG, Fassnacht M, Wiegering V (2023) Prognostic value of the Weiss and Wieneke (AFIP) scoring systems in pediatric ACC - a
mini review. Endocr Relat Cancer 30:e220259. https://doi.org/10. 1530/ERC-22-0259.
15. Wieneke JA, Thompson LD, Heffess CS (2003) Adrenal cortical neo- plasms in the pediatric population: a clinicopathologic and immu- nophenotypic analysis of 83 patients. Am J Surg Pathol 27:867-881. https://doi.org/10.1097/00000478-200307000-00001.
16. Magro G, Esposito G, Cecchetto G, Dall’Igna P, Marcato R, Gambini C, Boldrini R, Collini P, D’Onofrio V, Salfi N, d’Amore E, Ferrari A, Bisogno G, Alaggio R (2012) Pediatric adrenocortical tumors: mor- phological diagnostic criteria and immunohistochemical expression of matrix metalloproteinase type 2 and human leucocyte-associated antigen (HLA) class II antigens. Results from the Italian Pediatric Rare Tumor (TREP) Study project. Hum Pathol 43:31-39. https:// doi.org/10.1016/j.humpath.2011.04.016.
17. Dall’Igna P, Virgone C, De Salvo GL, Bertorelle R, Indolfi P, De Paoli A, Buffa P, Conte M, Esposito G, Inserra A, Candiotto C, D’Onofrio V, Boldrini R, Ferrari A, Bisogno G, Alaggio R, Cecchetto G (2014) Adrenocortical tumors in Italian children: analysis of clini- cal characteristics and P53 status. Data from the national registries. J Pediatr Surg 49:1367-1371. https://doi.org/10.1016/j.jpedsurg.2014. 03.006.
18. Chatterjee G, DasGupta S, Mukherjee G, Sengupta M, Roy P, Arun I, Datta C, Mishra PK, Banerjee S, Chatterjee U (2015) Usefulness of Wieneke criteria in assessing morphologic char- acteristics of adrenocortical tumors in children. PediatrSurg Int 31:563-571. https://doi.org/10.1007/s00383-015-3708-x.
19. Das S, Sengupta M, Islam N, Roy P, Datta C, Mishra PK, Banerjee S, Chaudhuri MK, Chatterjee U (2016) Weineke criteria, Ki-67 index and p53 status to study pediatric adrenocortical tumors: Is there a correlation? J Pediatr Surg 51:1795-1800. https://doi.org/ 10.1016/j.jpedsurg.2016.07.014.
20. Jehangir S, Nanjundaiah P, Sigamani E, Burad D, Manipadam MT, Lea V, Ly T, Holland AJA (2019) Pathological prognosti- cation of paediatric adrenocortical tumours: Is a gold standard emerging? Pediatr Blood Cancer 66:e27567. https://doi.org/10. 1002/pbc.27567.
21. Gupta N, Rivera M, Novotny P, Rodriguez V, Bancos I, Lteif A (2018) Adrenocortical Carcinoma in Children: A Clinicopathologi- cal Analysis of 41 Patients at the Mayo Clinic from 1950 to 2017. Horm Res Paediatr 90:8-18. https://doi.org/10.1159/000488855 ..
22. Paschoalin VP, Tucci Júnior S, G Estevanato A, B Tiraboschi R, Antonini SR, Muglia VF, Chahud F, Mermejo LM, de Bessa Júnior J, Fernandes Molina CA (2022) Histopathological Crite- ria for Paediatric Adrenocortical Carcinoma. Horm Res Paediatr 95:347-353. https://doi.org/10.1159/000524892.
23. Renaudin K, Smati S, Wargny M, Al Ghuzlan A, Aubert S, Leteurtre E, Patey M, Sibony M, Sturm N, Tissier F, Amar L, Bertherat J, Berthozat C, Chabre O, Do Cao C, Haissaguerre M, Pierre P, Briet C, Vezzosi D, Lifante JC, Pattou F, Mirallie E, Baudin E, Cariou B, Libe R, Drui D; for Comete-Cancer Network (2018) Clinicopatho- logical description of 43 oncocytic adrenocortical tumors: impor- tance of Ki-67 in histoprognostic evaluation. Mod Pathol 31:1708- 1716. https://doi.org/10.1038/s41379-018-0077-8.
24. Mete O, Gucer H, Kefeli M, Asa SL (2018) Diagnostic and Prog- nostic Biomarkers of Adrenal Cortical Carcinoma. Am J Surg Pathol 42:201-213. https://doi.org/10.1097/PAS.0000000000000943.
25. Angelousi A, Kyriakopoulos G, Athanasouli F, Dimitriadi A, Kassi E, Aggeli C, Zografos G, Kaltsas G (2021) The Role of Immunohistochemical Markers for the Diagnosis and Prognosis of Adrenocortical Neoplasms. J Pers Med 11:208. https://doi.org/ 10.3390/jpm11030208.
26. Tahar GT, Nejib KN, Sadok SS, Rachid LM (2008) Adrenocorti- cal oncocytoma: a case report and review of literature. J Pediatr Surg 43:E1-3. https://doi.org/10.1016/j.jpedsurg.2007.12.067.
27. Gumy-Pause F, Bongiovanni M, Wildhaber B, Jenkins JJ, Chardot C, Ozsahin H (2008) Adrenocortical oncocytoma in a child. Pediatr Blood Cancer 50:718-721. https://doi.org/10.1002/pbc.21090.
28. Lim YJ, Lee SM, Shin JH, Koh HC, Lee YH (2010) Virilizing adrenocortical oncocytoma in a child: a case report. J Korean Med Sci 25:1077-1079. https://doi.org/10.3346/jkms.2010.25.7.1077.
29. Kanitra JJ, Hardaway JC, Soleimani T, Koehler TJ, McLeod MK, Kavuturu S (2018) Adrenocortical oncocytic neoplasm: A sys- tematic review. Surgery. 164:1351-1359. https://doi.org/10.1016/j. surg.2018.04.044.
30. Kirkwood JM, Butterfield LH, Tarhini AA, Zarour H, Kalinski P, Ferrone S (2012) Immunotherapy of cancer in 2012. CA Cancer J Clin 62:309-335. https://doi.org/10.3322/caac.20132.
31. Giordano TJ, Berney D, de Krijger RR, Erickson L, Fassnacht M, Mete O, Papathomas T, Papotti M, Sasano H, Thompson LDR, Volante M, Gill AJ (2021) Data set for reporting of carcinoma of the adrenal cortex: explanations and recommendations of the guidelines from the International Collaboration on Cancer Reporting. Hum Pathol 110:50-61. https://doi.org/10.1016/j. humpath.2020.10.001.
32. Picard C, Orbach D, Carton M, Brugieres L, Renaudin K, Aubert S, Berrebi D, Galmiche L, Dujardin F, Leblond P, Thomas-Teinturier C, Dijoud F (2019) Revisiting the role of the pathological grading in pediatric adrenal cortical tumors: results from a national cohort study with pathological review. Mod Pathol 32:546-559. https://doi. org/10.1038/s41379-018-0174-8.
33. Juhlin CC, Bertherat J, Giordano TJ, Hammer GD, Sasano H, Mete O (2021) What Did We Learn from the Molecular Biology of Adrenal Cortical Neoplasia? From Histopathology to Transla- tional Genomics. Endocr Pathol 32:102-133. https://doi.org/10. 1007/s12022-021-09667-0.
34. Martins-Filho SN, Almeida MQ, Soares I, Wakamatsu A, Alves VAF, Fragoso MCBV, Zerbini MCN (2021) Clinical Impact of Pathological Features Including the Ki-67 Labeling Index on Diagnosis and Prognosis of Adult and Pediatric Adrenocorti- cal Tumors. Endocr Pathol 32:288-300. https://doi.org/10.1007/ s12022-020-09654-x.
35. Fay AP, Signoretti S, Callea M, Teló GH, Mckay RR, Song J, Carvo I, Lampron ME, Kaymakcalan MD, Poli-de-Figueiredo CE, Bellmunt J, Hodi FS, Freeman GJ, Elfiky A, Choueiri TK (2015) Programmed death ligand-1 expression in adrenocortical carcinoma: an exploratory biomarker study. J Immunother Cancer 3:3. https://doi.org/10.1186/s40425-015-0047-3.
36. Habra MA, Stephen B, Campbell M, Hess K, Tapia C, Xu M, Rodon Ahnert J, Jimenez C, Lee JE, Perrier ND, Boraddus RR, Pant S, Subbiah V, Hong DS, Zarifa A, Fu S, Karp DD, Meric-Bernstam F, Naing A (2019) Phase II clinical trial of pembrolizumab efficacy and safety in advanced adrenocortical carcinoma. J Immunother Cancer.7:253. https://doi.org/10.1186/ s40425-019-0722-x.
37. Raj N, Zheng Y, Kelly V, Katz SS, Chou J, Do RKG, Capanu M, Zamarin D, Saltz LB, Ariyan CE, Untch BR, O’Reilly EM, Gopalan A, Berger MF, Olino K, Segal NH, Reidy-Lagunes DL (2020) PD-1 Blockade in Advanced Adrenocortical Carcinoma. J Clin Oncol 38:71-80. https://doi.org/10.1200/JCO.19.01586.
38. Geoerger B, Kang HJ, Yalon-Oren M, Marshall LV, Vezina C, Pappo A, Laetsch TW, Petrilli AS, Ebinger M, Toporski J,
Glade-Bender J, Nicholls W, Fox E, DuBois SG, Macy ME, Cohn SL, Pathiraja K, Diede SJ, Ebbinghaus S, Pinto N (2020) Pem- brolizumab in paediatric patients with advanced melanoma or a PD-L1-positive, advanced, relapsed, or refractory solid tumour or lymphoma (KEYNOTE-051): interim analysis of an open-label, single-arm, phase 1-2 trial. Lancet Oncol 21:121-133. https://doi. org/10.1016/S1470-2045(19)30671-0.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.