Clinical Routine Application of the Second-generation Neuroendocrine Markers ISL1, INSM1, and Secretagogin in Neuroendocrine Neoplasia: Staining Outcomes and Potential Clues for Determining Tumor Origin

Staining of proteins associated to the protein machinery regulating the secretory granules of neuroendocrine cells has revolutionized endocrine pathology, starting with the Grimelius silver stain in the 1960s, followed by immunohistochemical analyses targeting chromogranin A (CGA) and synaptophysin (SYP). The diagnostic arsenal has since expanded, with ISL LIM homeobox 1 (ISL1), INSM transcriptional repressor 1 (INSM1), and secretagogin (SECG) to name a few novel markers exhibiting high sensitivity and specificity for neuroendocrine neoplasms (NENs).

ISL1 was originally found to be expressed in neuroendocrine cells as well as in neurons of the peripheral nervous system [1]. Since then, ISL1 has been established as a master DNA-binding transcriptional activator that is essential for pancreatic exo- and endocrine differentiation, and ISL1 has been found to bind directly to the insulin gene promoter and regulate insulin gene expression [2‐4]. Today, we know that ISL1 also carries ubiquitous roles in embryological processes, such as regulation of the BMP, WNT, and FGF signaling pathways in the development of external genitalia [5]. From a diagnostic perspective, ISL1 immunoreactivity was originally reported as a specific marker of pancreatic NENs (Pan-NENs), and the authors suggested that ISL1 immunohistochemistry could be employed to determinate the primary site of a metastatic NEN, in which a positive stain strongly would indicate a pancreatic origin [6]. In a succeeding publication, ISL1 immunoreactivity was also noted in duodenal NENs [7], followed by a report demonstrating ISL1 expression in NENs developed from extra-pancreatic locations, including all examined rectal NENs, and a majority of duodenal NENs, and in subsets of colonic NENs, appendiceal NENs, and small intestinal NENs (SI-NENs) [8]. Subsequently, ISL1 expression has also been reported as a highly sensitive marker also for additional neuroendocrine neoplasms, such as Merkel cell carcinomas, pheochromocytomas, paragangliomas, and medullary thyroid carcinomas [9]. As of this, the diagnostic role of ISL1 immunoreactivity has shifted somewhat, from an original viewpoint suggesting Pan-NEN specificity to that of a general neuroendocrine marker. Even so, ISL1 immunoreactivity could still indicate tissue-specific origin regarding metastatic well-differentiated NENs, in which the 3-marker panel TTF1, CDX2, and ISL1 could separate rectal and Pan-NENs from lung and SI-NENs [10]. ISL1 immunoreactivity has also been reported in poorly differentiated neuroendocrine carcinomas (NECs), suggesting that the marker could be of value in determining neuroendocrine differentiation also in dedifferentiated forms of the disease [9, 11].

INSM1 was originally described as an insulinoma-associated transcription factor that plays a key role in neurogenesis and neuroendocrine cell differentiation during embryonic development [12, 13] and has since been thoroughly investigated for its role as a marker of neuroendocrine differentiation. Studies have found that INSM1 is a reliable marker for neuroendocrine differentiation in various tumors, including Merkel cell carcinomas and NENs of the lung, gynecological tract, prostate, and head-neck region s[14‐21]. Notably, for pulmonary NECs, INSM1 has been proposed as a more sensitive marker than conventional neuroendocrine markers of the first generation [22].

SECG is a 32 kDa protein with calcium-binding properties discovered some 20 years ago, and initial analyses did pinpoint a selective expression of Langerhans cells in pancreatic tissues as opposed to the exocrine compartment [23‐25]. Subsequent analyses have revealed rather neuroendocrine-specific expression patterns, and the marker has been suggested to be a complement to the first-line marker CGA, as SECG was consistently positive also in colorectal NENs negative for CGA [26, 27].

At the Karolinska University Hospital, we have routinely used CGA and SYP immunohistochemistry as screening markers of neuroendocrine differentiation. Other markers such as CD56 and chromogranin B have been used much more periodically at our institution, as the former marker is fairly promiscuous in terms of tumor specificity, and the latter has a limited clinical usage in our experience—except for prognostication of pheochromocytomas and paragangliomas (PPGLs) [28, 29]. We introduced ISL1 immunohistochemistry in 2017 after successful validation of antibodies for clinical routine purposes, followed by INSM1 later the same year, while SECG was included in the diagnostic arsenal in late 2018. We here present the staining results of these newly established markers in a large NEN cohort, validating several previous findings and highlighting novel associations of potential diagnostic value. As most previous studies regarding these second-generation neuroendocrine markers are retrospective in nature, we wanted to share our experiences using a prospective staining setting in a high-volume referral center.

We present our institutional outcome of clinical routine staining for neuroendocrine markers of both first- (CGA, SYP) and second-generations (ISL1, INSM1, SECG), and verify previously suggested immunophenotypic patterns of clinical use. As our material rely solely on a prospectively stained material in which all markers were investigated as part of the routine clinical workup, we here conclude that the markers have adequate sensitivity and specificity when put against tumors with an initial histological suspicion of neuroendocrine differentiation, but in the end not fulfilling the criteria for a bona fide NEN. We also propose a comprehensive scheme regarding tissue-specific immunohistochemical profiles for diagnostic purposes, and suggest that ISL1, INSM1, and SECG can aid in the proper identification of the primary site of a metastatic NEN with unknown origin (Fig. 3).

In general, all three markers of the second generation seem to display high sensitivity for NENs, despite with some tissue-specific profiles (discussed below). ISL1, INSM1, and SECG all presented with exceedingly high PPVs concerning the distinction between NENs and non-NENs in the clinical setting, irrespective of WHO grade or tumor origin (primary tumor vs. metastasis). Hence, a focally positive or positive ISL1, INSM1, and/or SECG staining (in > 30% of cells, which was the cut-off for focal positivity in the NEN group) is highly indicative of a NEN when the pathologist is facing a non-NEN tumor with a histological suspicion of a NEN. Regarding the specificity of the markers in our series, SECG stood out as the most reliable performer in this aspect. Although only few of our institutional non-NENs were stained for this marker, our data could suggest that the excellent SECG specificity makes it a good candidate for clinical tumoral investigations in which a neuroendocrine differentiation is suspected.

From a clinical standpoint, one of the most pressing matters is to identify immunohistochemical markers with superior sensitivity compared with the expression of traditional markers of the first generation, which are recurrently reduced or lost in high-grade NECs. In our material, we verify that ISL1 is a sensitive marker also for the identification of NECs with an overall sensitivity of 86%. Although this was lower than the sensitivity for the first-generation marker SYP (100%), one should note that several of the ISL1 negative NECs cases in our series were derived from the lower GI tract (an anatomical region in which NENs recurrently are reported absent for ISL1 expression). Overall, the majority of NENs (including NECs) with positive ISL1, INSM1, or SECG expression were also positive for both CGA and SYP, but subsets of these NENs were CGA-negative while consistently SYP-positive—vaguely suggesting that a combination of SYP adjoined by second-generation neuroendocrine markers could constitute a highly sensitive and specific panel for clinical usage (Supplementary Table 3).

Several tissue-specific patterns were recognized in this study, of which some have been proposed in previous publications. For example, positivity for CGA, SYP, ISL1, INSM1, and SECG was a common profile for Pan-NENs irrespectively of tumor grade or tumor site (primary lesion vs. metastatic lesion), while an immunohistochemical profile consisting of a negative ISL1 staining adjoined by positivity for other neuroendocrine markers of both the first- and second-generation would strongly suggest an origin in the small intestine or appendix. Indeed, previous studies have pinpointed high sensitivity for ISL1 and INSM1 immunoreactivity in Pan-NENs, as well as widespread ISL1 negativity in SI-NENs [6, 8, 31‐33].

A few previously uncharacterized profiles of potential value for diagnostic purposes were also evident. For colorectal NENs, positivity for ISL1, INSM1, SECG, and SYP was a consistent profile, even in the absence of CGA immunoreactivity—the latter phenomenon not being uncommon for NENs arising in the large intestine (Table 1, Fig. 3). To our knowledge, INSM1 and SECG immunoreactivity has not yet been thoroughly assessed in colorectal NENs apart from single observations in limited case series [26, 34]. Moreover, rare cases of renal NENs and NENs arising in teratomas included in our study were consistently positive for ISL1, which are novel findings worth exploring in larger materials.

For PPGLs, a profile with positive ISL1 and INSM1 stainings adjoined by negative immunoreactivity for SCG was commonly observed (Table 1, Fig. 3). These results are somewhat contradictory from previous findings suggesting widespread SCG positivity in NENs from the adrenal glands [24, 26, 27]. However, we validated our findings by also scrutinizing the immunoreactivity in adjacent, histologically normal adrenal glands from the same patients (Supplementary Fig. 1). We can only speculate regarding the underlying reasons for these methodological discrepancies, but previous studies were either conducted using northern blot analyses [24] or with immunohistochemistry using polyclonal antibodies for the majority of analyses [26, 27], whereas we employed a monoclonal SCG antibody. Future studies regarding SCG expression in normal adrenal medulla and PPGLs are therefore highly warranted, as metastatic PPGLs can be hard to pinpoint, as established clinical markers (such as GATA3 and S100) not always stain positive, making an auxiliary high-sensitivity marker of imminent clinical value.

This study mainly revolves around ISL1 immunohistochemistry, which is due to the fact that the marker was introduced in our clinical screening well before INSM1 and SECG. As the aim of this study was to report the clinical utility of all these markers, there is a predominance of ISL1 stained tumors that often lacked stains of the two other markers. As tumors were not stained in retrospect, we recognize that our conclusions regarding INSM1 and SECG immunoreactivity as discriminative markers should be interpreted with caution and our findings reproduced in larger materials. Moreover, many of our NEN subgroups are rare manifestations of NENs not commonly encountered in clinical practice, for example renal, esophageal and cervical NENs as well as NENs in teratomas. Although our ISL1, INSM1, and SECG data regarding many of these NEN subtypes are novel and therefore warrant attention, we lack power to safely assess the immunohistochemical profiles of these extraordinary NEN subtypes from a diagnostic standpoint.

To summarize, we verify previous observations that immunohistochemical staining using neuroendocrine markers of the second generation should act as a complement to the established markers CGA and SYP and recommend these markers to be implemented in routine clinical practice.