Transcription Factor Immunohistochemistry in the Classification of Pituitary Neuroendocrine Tumor/Adenoma: Proposal in a Limited-Resource Setting

Histologically confirmed pitNET/PAs diagnosed between January 2018 and April 2021 were included in this study. A total of 472 pitNET/PAs were examined, among which 116 cases had to be excluded because of extensive apoplexy, unavailability of tissue block, insufficient tumor tissue, or a revision of the original diagnosis. The resultant 356 tumors were subjected to the following panel of antibodies: PIT1 (1:500; Novo Biological), TPIT (1:500; Atlas CL6251), SF-1 (1:300; Abcam 217317), adrenocorticotropin (ACTH; 1:70; Dako 02A3), thyroid-stimulating hormone (TSH; 1:50; Dako 0042), prolactin (PRL; 1:200; Dako polyclonal), follicle-stimulating hormone (FSH; 1:50; Dako C10), luteinizing hormone (LH; 1:50; Dako C93), growth hormone (GH; 1:400; Dako hGH), and Ki-67 (1:300; Cell Marque SP6). A tissue microarray (TMA) was constructed for 158 cases, and the rest (198; 55.7%) underwent whole slide evaluation. Table 1 summarizes the names, abbreviations, and definitions of the various types of adenomas referred to in the text. TMAs were constructed by punching 3 cores of 2 mm diameter each from formalin-fixed, paraffin-embedded blocks, which were then inserted into a recipient paraffin block using an automated tissue microarrayer (Quick Ray Master UATM-272B). A hematoxylin-eosin–stained slide of each TMA block was generated to assess TMA quality. IHC was performed on a Ventana Biotech automated system with appropriate positive and negative controls run concurrently. PTF and hormone positivity was defined by a cutoff of 5% staining in tumor cells.⁵  Intensity of staining was recorded using the Allred score as no (0), weak (1+), moderate (2+), and strong (3+) staining and was added to a proportion scoring (0, no staining; 1, 1%; 2, >1%–10%; 3, 11%–33%; 4, 34%–66%; 5, ≥67% of tumor cells, respectively). The patient records were analyzed for clinical, biochemical, and radiologic features. If the hormonal levels were higher than the sex-adjusted limits, or if stigmata of acromegaly, Cushing disease, hyperprolactinemia (which could not be explained by stalk effect), and hyperthyroidism were present, then the tumor was labeled as a functioning PitNET/adenoma (FPT). The disease/recurrence-free survival (DFS) was calculated based on follow-up hormone level and neuroimaging.

This study was carried out after appropriate clearance by the institution’s ethics committee.

Statistical analysis was performed using SPSS software (version 23, IBM Corp, Armonk, New York). The results were analyzed using the χ² and Fisher tests. DFS was calculated from the date of diagnosis to the date of recurrence or detection of residual disease. Kaplan-Meier survival curves were used to estimate overall survival, and the log-rank (Mantel-Cox) test was used to assess differences in survival between groups. Differences with P ≤ .05 were considered statistically significant.

The mean age of patients included in this cohort was 42 years, with a male-to-female ratio of 1.03:1. Based on clinical features and biochemical profiles, 203 of the 356 cases (57%) were classified as nonfunctioning tumors (NFTs). The individuals with NFTs were on average a decade older and more frequently of the male sex than those with FPTs (n = 153 of 356, P < .001). One case of gonadotroph pitNET (GT) was nonfunctional at the time of diagnosis; however, on follow-up it became functional. Tumor size ranged between 0.36 and 6.30 cm, with an average of 2.73 ± 1.51 cm. The mean size of FPT (2.2 ± 1.59 cm) pitNETs was significantly smaller than that of the NFTs (3.19 ± 1.28 cm; P = .03).

On application of hormone IHC, 224 of the total 356 pitNET/PAs (62.9%) were hormone IHC positive. These were classified as: 65 of 356 were somatotroph pitNETs (18.3%), 53 of 356 were corticotroph pitNETs (CTs; 14.9%), 41 of 356 were GTs (11.5%), 31 of 356 were mammo-somatotroph tumors (8.7%), 12 of 356 were lactotroph pitNETs (3.4%), and 10 of 356 were thyrotroph pitNETs (2.8%). A total of 12 of 356 tumors (3.4%) expressed more than 1 hormone IHC and were called plurihormonal pitNETs (PHTs). The remaining 132 of 356 cases (37.1%) were hormone-negative pituitary neuroendocrine tumors (HNPTs). Among the NFTs (n = 203), 122 (60%) were hormone negative as well, and 81 (40%) expressed hormone IHC (silent pitNET/PAs).

On applying PTF IHC, 336 of 356 cases (94.4%) showed PTF positivity. A total of 20 of 356 tumors (5.6%) were negative for the 3 PTFs and were labeled as triple-negative pituitary neuroendocrine tumors. Isolated nuclear staining for SF-1, PIT1, and TPIT was seen in 151 of 356 (42.4%), 121 of 356 (34%), and 57 of 356 (16%) pitNETs, respectively. More than 1 PTF was expressed in same the tumor cell population in 7 of 356 cases (2%).

Cohen κ was run to determine the agreement between “integrated diagnosis” and the isolated clinical, hormone IHC–based, and PTF IHC–based classification. The highest agreement was achieved using a combination of hormone and PTF IHC (κ = .988, P < .001). There was moderate agreement between the final integrated diagnosis and only pituitary hormone IHC (κ = .592, P < .001) and only PTF IHC (κ = .570, P < .001) diagnoses (Table 2). A summary of the final subtypes of pitNETs based on clinical, biochemical, histopathologic, and IHC features is depicted in Figure 1.

The clinical features of FPT, along with corresponding histopathology, have been elaborated in Table 3. Clinical endocrine status, serum hormone levels, and hormone expression within tumors were reasonably correlated in most pitNETs. Among the 153 FPTs, 4 (2.6%) were classified differently from their clinical presentation. One patient with weight gain and Cushingoid habitus had a diagnosis of ST, and another patient with coarseness of facies was ACTH and TPIT positive on pathology. Even though the vast majority of GTs were clinically silent, 2 cases had dissimilar functional presentations. One patient with GT presented with markedly elevated serum prolactin (92.33 ng/mL) levels, whereas another, albeit with NFT, presented with mildly elevated ACTH levels (70.10 pg/mL), that is, whispering CT.

SF-1 expression was appreciated in 106 cases (80.3%), whereas 3 (1.48%) showed TPIT and 8 (6.1%) adenomas showed PIT1 staining, respectively. The last of these were labeled as immature PIT1-lineage PitNETs (IPLTs). Only 15 of the 132 HNPT cases (ie, 11.4% cases overall) were true null cell pitNET (NCT), being negative for all 3 PTFs also. These cases were positive for cytokeratin and negative for GATA3, PTF, and all hormone IHCs. Internal control for PTF and hormone IHC was positive in all these cases.

PHTs were defined by exhibiting discordant lineages, that is, 2 or more lineage hormones (n = 8; 53.3%) or 2 or more lineage PTFs (n = 7; 46.6%) expression, with the exception of GH/PRL/TSH or LH/FSH coexpression. The most common PTF staining was that of SF-1, seen in 10 of the 15 PHTs (66.6%; Figure 2). Table 4 gives the clinical and pathologic aspects of these cases. The possibility of trapped nontumorous tissue was excluded, as was the presence of double and multiple synchronous adenomas. Two patients (2 of 356; 0.5%) who showed coexpression of GH, PRL, and TSH along with diffuse nuclear staining for PIT1 (Figure 3) were categorized as having “mature plurihormonal PIT1 lineage pitNET/adenoma (MPPT)” as per the 2022 WHO Classification for Endocrine and Neuroendocrine tumors.⁶ 

A total of 124 of 356 tumors (34.8%) were recategorized after PTF IHC. Most (94.3%) belonged to the hormone IHC–negative category (117 of 124 reclassified cases). Among the HNPTs, 106 GTs, 8 IPLTs, and 3 CTs received a diagnosis of PTF. Other than the HNPTs, 3 GH-positive pitNETs and 1 tumor each expressing ACTH and TSH stained for more than 1 PTF and were subsequently reclassified as PHT. Two patients who had PHT on hormone IHC, because they expressed GH, PRL, and TSH, were PIT1 positive, and hence the integrated diagnosis was revised to MPPT.

Most pitNETs of this lineage were positive for both hormone IHC and PIT1. Three patients who were positive for GH alone or with PRL were negative for PIT1 (as well as SF-1 and TPIT). Hence, the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of GH in detecting somatotroph lineage were superior to those for PIT1 (Figure 4). However, PIT1 fared better than PRL and TSH, with higher sensitivity, specificity, PPV, and NPV. Overall, 8 of 121 PIT1-positive patients (6.6%) were negative for hormone IHC, diagnosed as IPLT. These cases were reanalyzed with GH, PRL, and TSH on whole section for confirmation of negativity. Another 2 patients had PIT1 and GH diffusely positive with focal TSH and PRL staining (Figure 3) categorized as MPPT.

Coexpression of TPIT and ACTH was seen in 50 CTs (90.1%), with a strong correlation between their Allred scores (Pearson r = .954). However, the sensitivity and specificity, as well as PPV and NPV, of TPIT were higher than those of ACTH in accurately diagnosing CT.

Most of the GTs (106 of 147; 72%) were positive for SF-1 only and were negative for both LH and FSH. The sensitivity and NPV of SF-1 were observed to be 100% (Figure 4). SF-1 and PIT1 showed diffuse and strong staining with significantly higher mean Allred scores compared with hormone IHC (P < .001). TPIT and ACTH did not differ in staining intensity or percentage positivity, and both showed focal staining in the majority.

Only 1 patient died during the course of follow-up—due to surgery-related complications and unrelated to the tumor—and hence was excluded from the analysis. The median DFS of the 296 patients where follow-up was available was 762 ± 151 days (25.4 ± 5 months). Among the broad subtypes, the shortest DFS was observed for NCTs (mean, 214 ± 44 days; 7.0 ± 1 months) followed by thyrotroph pitNETs (mean, 401 ± 102 days; 13.4 ± 3.5 months). In contrast, the longest DFS was observed in the GTs, with a median DFS of 49.5 months (1506 days). Among the 8 IPLTs, 3 recurred after 594 days (median duration) of follow-up. Also, female sex (log rank, P = .01); functional status (log rank, P = .01); and microscopic features of aggressiveness, such as mitosis more than 2 per 10 high-power fields, nuclear pleomorphism, nuclear multinucleation, Ki-67 index greater than 3%, and mucosa or bone infiltration (log rank, P = .01), were associated with significantly poorer DFS. The MPPTs did not differ in terms of radiologic or microscopic features of aggression from IPLTs (P = .81) and PHTs (P > .99). Cases expressing more than 1 PTF had more frequent recurrences than those that expressed only 1 PTF; however, this difference was not statistically significant (P = .06). No recurrences were seen in the 2 cases of MPPTs (mean duration of follow-up of 52 days), whereas 2 PHTs experienced recurrences after a mean follow-up of 64 days.

This is the largest surgical series from India where PTF IHC has been used in conjunction with hormone IHC to classify 356 pitNETs/adenomas. An adequate categorization of these tumors is clinically significant because the various pitNETs differ in their biologic behavior and response to treatments. The WHO classification of pitNETs was updated in 2022, and many new/redefined entities, such as NCT, MPPT, and IPLT, were introduced that could not be diagnosed without applying PTFs.⁶  However, the incorporation of these antibodies into daily practice faces several shortcomings, mainly pertaining to the lack of consensus on establishing cutoffs for these antibodies.^7,8 

Clinically incongruous pitNETs cases in the present study can be explained by an overlap of many of the signs and symptoms of acromegaly and Cushingoid habitus with obesity, metabolic syndrome, diabetes mellitus, etc.⁹  Also, the stalk effect by sellar and suprasellar masses, including pitNETs, is a known phenomenon that can lead to the misdiagnosis of lactotroph-secreting tumor on clinical or biochemical grounds. Similar findings were observed in a single case in the present series that expressed SF-1 diffusely while being negative for PRL and PIT1. Change in hormone secretion by pitNETs is an extremely rare and poorly understood phenomenon. Andino-Ríos et al¹⁰  have reported a similar case and attributed it to changes in gene expression pattern. In the current series, only 1 patient (0.3%; 1 of 356) changed from NFT to functional GT pitNET.

Most of the studies on PTF have used SF-1 and/or PIT1 in the setting of NFT.^4,11  Three of the largest published studies, by Silva-Ortega et al⁷ , Mete et al,⁸  and Lenders et al,¹²  have used all 3 PTFs (ie, SF-1, TPIT, and PIT1), analogous to our study. Silva-Ortega et al⁷  analyzed 146 pitNETs and successfully reclassified 120 (82%) of the HNPTs. In the current study, this number was slightly higher (127 of 132 cases; 88.6%). The predominant pitNET/adenoma type was GT (41.3%), similar to the others.^7,8,12 

PTF was indispensable in diagnosing 106 of 147 GTs (72%), and 3 of 55 cases of CT (5.5%). Previously published studies also detected SF-1 positivity in two-thirds of HNPTs. In contrast, Nishioka et al¹¹  reported lower rates of SF-1–only positive GTs (79 of 379; 20.8%) and higher TPIT-only positive CTs (32 of 83; 26.9%), which exceeds immature PIT1 lineage HNPTs (2 of 48; 1.7%).¹¹  In the current study, PIT1 stained 8 of 132 HNPT cases (6.1%), which exceeded the number of TPIT-positive hormone-negative silent CT cases (3 of 132; 2.3%). A similar occurrence was observed in a study by Lee and colleagues.¹³  These variations in frequency can be explained by (1) a difference in antibody clones, as well as (2) a higher cutoff for defining positivity (80% by Nishioka et al¹¹). Similarly, Lenders et al¹²  in a series of 171 pitNETs reclassified 12% of tumors after PTF IHC.¹² Table 5 compares the distribution of pitNET types in these studies.

Complete absence of all hormone and PTF IHC was seen in 4.2% (15 of 356) of all pitNETs and 11.4% (15 of 132) of HNPTs qualifying for NCT. IPLTs differed from NCT by virtue of PIT1 expression; however, they contrasted in their mean DFS. This highlighted the importance of performing PIT1 immunostain because median DFS for NCT was 7 months (214 days), whereas it was 97 months (819 days) for IPLT (Kaplan-Meier, log rank P = .02). However, the follow-up analysis of the series is quite short. This is a significant caveat of the recurrence and DFS data because pitNETs may recur several years or decades after initial presentation.

The PTF expression in hormone IHC–negative PitNETs suggests an origin from primitive cells. It is postulated that the neoplastic cells underwent a maturation arrest before reaching adequate maturity for hormone production. A study detecting PIT1 mRNA by Yamada et al¹⁴  demonstrated its presence bereft of GH, PRL, or TSH transcripts in 3 NFTs. Dedifferentiation or suppression of the hormone transcripts can be another mechanism. PTF-positive hormone-negative tumors were addressed in the Armed Forces Institute of Pathology fascicles of PA; however, no unified nomenclature exists for the same. There is an evolution in the terminology of certain pitNETs, such as the PIT1-positive adenomas that lack terminal differentiation, referred to in the latest WHO endocrine and neuroendocrine tumors classification (2022) as immature PIT1 lineage pitNET/adenomas and by the previous edition of WHO as plurihormonal “poorly differentiated” PIT1-positive adenoma and silent subtype 3 adenoma.^6,15  Because many tumors of PIT1 lineage can be treated medically, PIT1 becomes indispensable in those cases that are nonfunctional and hormone IHC negative.

Coexpression of PIT1 with SF-1 and PIT1 with TPIT was seen in 7 of 356 cases (1.96%), which were together categorized as PHT. These tumors have been reported in 0.9% to 3.2% of resected pitNETs, and most of these correspond to PIT1 and corticotroph lineage.^15,16  Micko et al¹⁷  observed coexistence of gonadotroph-PIT1 lineage hormones or PTF in 20 cases (4.4% of 458 PAs). In this series, 7 of the 15 PHTs (4.7%) were PIT1 positive. Tordjman et al¹⁸  have described 1 case of a collision double adenoma in a pregnant woman that expressed PIT1-SF1 and GH, TSH, and FSH, along with α subunit. In our study (n = 356), there were 5 cases (1.4%) of corticotroph-PIT1 lineage, 9 cases (2.5%) with PIT1-GT lineage, and 2 cases (0.6%) where SF-1 was expressed with ACTH and GH (ie, evidence of all 3 lineages). The last “tri-lineage pitNET/adenomas” highlight the need to expand the definition/nomenclature of plurihormonal pitNET to include “pluri-lineage” or “multi-lineage” as a more comprehensive terminology and one that is in keeping with current lineage-based classification. The most extensive series on PHT by Micko et al¹⁷  labels these cases as plurihormonal adenomas with unusual immunohistochemical combinations. The PHTs with more than 1 PTF expression also had a short mean DFS of 735 days (median, 563 days).

Recent updates have addressed a much-debated distinction between the plurihormonal tumors expressing discordant hormone and transcription factor and those that express PIT1 with GH, PRL, and TSH combinations (ie, MPPT).⁶  The previous 2 WHO classifications have also rebranded the latter, going from calling them silent subtype 3 adenomas to plurihormonal PIT1-positive adenomas in 2017.^6,19  Because the latter is not of unexpected or unusual lineage, this revelation provides some clarity. Despite this, the authors would like to bring attention to the fact that although the 2022 WHO classification dropped the term “plurihormomal” from the IPLT, it is still part of the mature PIT1 pitNET.⁶  Both of these are categorized under the pitNETs of PIT1 lineage.

The current panel for classification includes the 6 adenohypophyseal hormones and available transcription factors. These antibodies, together with Ki-67 and p53, reach a dozen. GATA-2, GATA3, α-estrogen receptor (α-ER), α subunit, and neuroD1 are other PTFs that have also been added to the list. This large panel is time-consuming and expensive, especially for those laboratories facing a high workload and having limited resources. A constant drawback in routine practice is the lack of clinical details on pathology forms, especially regarding functional status or hormone levels. Also, most cases in our practice that undergo surgical excision are clinically nonfunctioning. The recommended approach is to perform the complete panel of 9 stains, costing approximately USD $55 (rupees 500) per case, in our setup. Recently published recommendations on the required ancillary tools of pitNET diagnosis and classification state “that a tiered approach IHC is recommended starting with PTF and using relevant hormones for the PTF identified.”²⁰  Performing immunostains in a sequential manner saves tissue and resources but can lead to a slight delay in the final reporting of cases by ∼2 days. Thus, by using 1 PTF for each lineage, followed by hormone IHC of the lineage, identified rare combinations can be picked up as well. However, the authors do emphasize the importance of careful evaluation of the patient’s clinical and laboratory data for accurate classification.

In our study, the antibodies for SF-1, PIT1, and TPIT yielded a higher specificity and sensitivity than most of the hormone IHC (Figure 4). An important question is whether PTF should proceed, follow, or substitute the hormone IHC. We propose to begin with a 3-antibody panel comprising SF-1, PIT1, and 1 immunostain for corticotroph lineage of either TPIT or ACTH, both of which showed similar performance in our cohort. McDonald et al⁴  have successfully used a combination of ACTH, PIT1, and SF-1 as screening markers and demonstrated excellent agreement between the diagnosis obtained via algorithm and the “integrated diagnoses.”⁴ 

As per our recommended algorithm, most tumors, ∼200 (or 56.1%), that were dominated by GTs and CTs could be diagnosed correctly with the PTF IHC panel as the initial step itself (Figure 5). This means that more than 50% of cases would be diagnosed accurately by using SF-1 and TPIT only. One-third of cases (121 of 356; 34%) that were only PIT1 positive would additionally require a second panel of immunostains for GH, PRL, and TSH (ie, a total of 6 IHC markers). What if we were to skip the PIT1 stain altogether and perform GH, PRL, and TSH along with SF-1 and TPIT? This would have led to the underdiagnosis of the 1.4% to 6% of IPLTs that are only PIT1 positive. The latter often have an aggressive prognosis compared with NCT. Also, STs and lactotroph pitNETs require performing one of the keratin antibodies (as a third panel) in the form of CAM5.2, AE1/AE3, or CK8/18 for distinguishing between the densely and sparsely granular types. Thus, we agree with the observations of Lee et al¹³  and reiterate the importance of performing PIT1, which cannot be substituted with GH, PRL, and TSH hormone IHC.

Thus, there were only 27 of 356 cases (7.6%) that were either negative for all 3 PTFs or expressed more than 1 PTF and would require a further hormone IHC panel to diagnose NCT or PHT. The complete panel of 9 antibodies per case for the average number of ∼165 cases annually at our tertiary care center would cost the lab approximately USD $10 000 (rupees 7.5 lakhs). However, as per our proposed algorithm, the cost per year would be USD $4250 (rupees 3.5 lakhs), reducing the cost annually by 50%, with a simultaneous gain in diagnostic accuracy. Clinical input of functional status is essential because these cases require a complete biochemical profile along with testing with all the available hormone IHC; however, this should include ACTH, GH, and PRL as a bare minimum. If all 3 PTFs are negative, that is, a triple-negative pituitary neuroendocrine tumor, a cytokeratin stain can be carried out (before hormone IHC), which, if positive, distinguishes them from paragangliomas. In this situation, GATA3, because of its positivity for most gonadotroph and thyrotroph adenomas as well as paragangliomas and metastases, is not beneficial.^21,22  Because of the fact that most GTs are negative for FSH and LH if PTFs are not used, GATA3 positivity alone cannot be used to differentiate them from paragangliomas. Thus, if PTFs are not used, only GATA3 positivity alone cannot distinguish a paraganglioma from a GT, because most of them are negative for FSH and LH.

Thus, ∼98% of the adenoma/pitNETs would be diagnosed correctly by this algorithm using hormone antibodies in a sequential manner where required (Figure 5). A total of 6 of 356 cases (1.7%) of PHT diagnosed because of a discordance between the hormone and PTF IHC lineages could potentially be missed. However, these pitNET/adenomas did not differ from the PHTs in terms of DFS, radiologic invasion, or microscopic aggression.

The diagnosis of pitNETs has evolved and requires PTF as a subclassification tool, which is now readily available and validated in most labs. Most of the studies using PTF have focused on the incidence and prevalence of these individual tumor types. We correlated the various subtypes of pitNETs with clinical and radiologic features and proposed an algorithm to accurately diagnose these tumors using the immunostains sequentially. PTF expression alone is not sufficient to initiate clinical treatment, although 87.7% of the NFTs expressed only PTF, making their application essential. We also conclude that PTF can simultaneously enhance classification and prognostication accuracy, diagnose new entities, and reduce the cost of analysis. A limitation of this study was that IHC for GATA3 and α-subunit was not performed, the latter because of nonavailability.

SF-1 and PIT1 should be included in the routine panel for guiding the classification. PIT1 positivity needs to be followed by hormone IHC, especially in nonfunctional cases. TPIT and ACTH can be used interchangeably as per availability of the laboratory. NCT phenotype correlates with the shortest DFS, highlighting the importance of performing all PTFs. In the PTF-negative and “pluri-factorial” adenomas requiring hormone IHC, the primary panel should include ACTH, PRL, and GH. More comprehensive terminologies are needed for these tumors.