The Ki-67 antigen in the new 2021 World Health Organization classification of lung neuroendocrine neoplasms.

Prof. Rosai's work has permeated the surgical pathology in many fields, including the 2017 World Health Organization classification on tumors of endocrine organs and pulmonary neuroendocrine cell pathology, with stimulating contributions which have also anticipated the subsequent evolution of knowledge. Among the many studies authored by Prof. Rosai, we would like to recall one of which whose topic has been encased in the new 2021 World Health Organization classification on lung tumors. This is an eminently practical paper dealing with the use of the proliferation antigen Ki-67 in lung neuroendocrine neoplasms. While these neoplasms are primarily ranked upon histologic features and Ki-67 labeling index does not play any role in classification, diagnostic dilemmas may however arise in severely crushed biopsy or cytology samples where this marker proves helpful to avoid misdiagnoses of carcinoids as small cell carcinoma. Another application of Ki-67 labeling index endorsed by the 2021 World Health Organization classification regards, alongside mitotic count, the emerging recognition of lung atypical carcinoids with increased mitotic or proliferation rates, whose biological boundaries straddle a subset of large cell neuroendocrine carcinoma. This article focuses on these two practical applications of the proliferation marker Ki-67 in keeping with the 2021 World Health Organization classification, which provides standards for taxonomy, diagnosis and clinical decision making in lung neuroendocrine neoplasm patients.

Introduction

Prof. Juan Rosai’s commitment in the domain of thoracic pathology has been outstanding and continuous over time, with pioneering definitions of new tumor pathology entities and clarifying descriptions of unusual tumor associations. Just to give some instances, we could simply enumerate his seminal contributions to the development and clarification of several thoracic pathology issues, such as: a) first description of the malignant small cell tumor of the thoracopulmonary region in childood, the so-called Askin-Rosai tumor, currently classified as belonging to the Ewing family of tumors with variable degrees of neuroectoderm differentiation [1]; b) first association between neuroendocrine neoplasms (NENs) of the thymus and MEN1 syndrome [2,3]; c) terminology definition and classification criteria on the thymus [2,4-8] and larynx [9] NENs, including the description of the spindle cell variant of thymic atypical carcinoid (AC) [10], neuroendocrine differentiation in thymic carcinoma [11] and neuroblastoma in adult thymuses [12]; d) editorial boarding responsability in the 1999 (2nd edition) and 2017 (4th edition) World Health Organization (WHO) histological typing of the thymus [13] and endocrine organ tumors [14], respectively; e) evolutionarily outlook on the origin and development of neuroendocrine cells and related neoplasms [15], with a scholar essay on the neural crest saga [16]; f) demonstration of Ki-67 labeling index as a managerial biomarker to avoid overdiagnosing carcinoids as small cell carcinoma in biopsies samples [17]; g) first description of desmoplastic small round cell tumor of the pleura, a type of sarcoma with multilinear lineage including neuroendocrine differentiation [18]; h) identification of florid vascular proliferations in high-grade neural and neuroendocrine neoplasms as a diagnostic clue [19], including small cell lung carcinoma (SCLC) [17]; i) immunohistochemical characterization of neuroendocrine markers, whether nuclear (i.e., Hu proteins) [20] or cytoplasmic (i.e., neuron specific enolase) [21] and their diagnostic utilization in the setting of NENs; j) association of typical carcinoid (TC) with Fechner’s acinic cell tumor of the lung [22]; k) follicular dendritic cell tumors arising in diverse anatomical sites, including mediastinum [23,24]; l) the impact of WHO classification of NENs, including those arising in the lung, to select an appropriate treatment [25]; m) one example of diffuse pulmonary neuroendocrine cell hyperplasia [26]; and n) the occurrence of metastatic pulmonary carcinoids to the thyroid featuring increased proliferation as the first clinical manifestation, which simulated medullary carcinoma [27].

Juan Rosai’s scientific genius, methodological rigor and encyclopedic culture, in a word his outstanding natural talent, have been able to juggle apparently unrelated topics, devise new diagnostic algorithms, create innovative classifications and provide biological interpretations of great scientific value. His seminal contributions have been resonating over time in the WHO classifications of thoracic tumors, including some echos even in the latest 5th 2021 edition, which is largely centered on precision medicine and multidisplinary teaming [28] whose objectives may be traced back in the work of Prof. Rosai as pathologist, scientist, mentor and teacher [29].

In this article dedicated to his memory and activity in international classifications, we would like to briefly hint at one topic of practical value in daily practice, which also appears in the new 5th edition of WHO classification on thoracic tumors [28]. This is embodied by the role of Ki-67 antigen immunostining in separating carcinoids from SCLC in biopsy samples. Since the theme of differential diagnosis has always been a hallmark of Prof. Rosai’s work, this approach seemed to us a respectful way to commemorate him by starting from one of his own original papers in the field of lung NENs, whose principles have been maintained in the 2021 WHO classification. A direct evolution of such a diagnostic role of Ki-67 antigen in lung NENs can also be perceived in the emerging concept of atypical carcinoids with increased mitotic or proliferation rates, which opens unexpected horizons to our understanding of these neoplasms.

The role of Ki-67 staining in lung neuroendocrine neoplasms

ROSAI’S CONTRIBUTION

In 2004, Prof. Rosai and one of us (GP) published as leading author a case series of seven preoperative fiberoptic bronchoscopic carcinoid biopsies, which had been originally misdiagnosed as SCLC [17]. These discordant diagnoses on biopsy samples were selected from all surgically resected carcinoids diagnosed over a 12-year period at the European Institue of Oncology (Milan, Italy) or obtained from the consultation files of Prof. Rosai. Furthemore, bronchial biopsies of nine consecutive patients with a clinically confirmed diagnosis of SCLC were used as a control group for histologic and immunohistochemistry (IHC) comparison. The starting idea was that Ki-67 antigen, a well known marker of cell proliferation in neuroendocrine pathology [30], would prove superior in the differential diagnosis between carcinoids and SCLC when used as single-shot marker as compared to other tumor descriptive characteristics, such as the anatomical location or other IHC markers. To verify this hypothesis, tumor position in the lung, histologic features and decoration for cytokeratins, chromogranin A, synaptophysin and TTF1 were analyzed for comparison. Microscopic examination of carcinoid biopsies (taken from central or peripheral tumors), showed extensive crush artifacts in 50% or more of the tissue fragments in four of the seven cases under evaluation, which hampered ready recognition of SCLC-hallmarking details, such as mitotic figures, necrosis and nuclear molding. The IHC study showed variable expression of epithelial and neuroendocrine markers, whereas TTF1 was the only one to by far prevail in SCLC as one would expect [31]. Ki-67 labeling index averaged 4.6% in carcinoid biopsies and 10.6% in paired surgical specimens (this difference was not statistically significant), while as many as 81.2% in the nine SCLC biopsies under evaluation. Appreciation of Ki-67 was readily apparent in severely crushed areas, where the fine inspection on cell size, chromatin pattern, necrosis and mitotic figures was challenging. The conclusion of the study was that Ki-67 labeling index was very useful for distinguish carcinoids with crush artifact from SCLC. In this study, Ki-67 labeling index was low-to-moderate (up to 20%) in carcinoids and highly expressed (50% or more) in SCLC [17].

WHO CLASSIFICATION OF LUNG NENS

The role of Ki-67 labeling index in the spectrum of lung NENs was recently been reviewed by us [32] and only its diagnostic application will be herein recalled according to the indications of the 2021 WHO classification [28]. Functionally speaking, Ki-67 antigen is deemed to play roles in both interphase G1 and mitosing cells, where its cellular distribution dramatically changes over cell cycle progression [33,34], while completely and rapidly (few hours) disappearing in G0 phase [35]. In interphase G1 cells, Ki-67 is at low levels and is required for heterochromatin and nucleolar organization [36], while during the cell cycle dramatically increases because it is essential to ribonucleoprotein sheath formation coating condensed mitotic chromosomes as a kind of surfactant to prevent aggregation [37-39] by controlling liquid-liquid phase separation of nucleolar proteins and RNAs [40]. This different distribution of Ki-67 antigen according to the dynamics of the cell cycle in normal and neoplastic cells results in diverse nuclear patterns upon IHC [41], which are not yet practiced in tumor pathology where all staining levels and expression patterns are counted to quantify the relevant labeling index [32]. The function of Ki-67 in tumor cells, beyond its diagnostic or prognostic effects depending on different tumor context, could also be exploited as an exit target for cancer therapy by variably interfering with its own multifaced biological activity [42].

The classification of lung NENs has been endorsed by the new 2021 WHO edition according to a conceptually unifying spectrum of lesions [28]. Accordingly, they comprise TC, AC, large cell neuroendocrine carcinoma (LCNEC) and SCLC, whose defining criteria are based on mitotic count per 2 mm2, necrosis assessment and a constellation of cytological and IHC traits for epithelial and neuroendocrine markers (Tab. I). Carcinoids exhibit medium-sized cells with polygonal to spindle shape, whereas SCLC and LCNEC are hallmarked by small-sized and large-sized cells, respectively. Paraneoplastic syndromes are most frequently seen in SCLC but can also be observed in carcinoids, whereas combined variants are prerogative of neuroendocrine carcinomas (NECs) (e.g., with adenocarcinoma or squamous cell carcinoma) and just anecdotal in carcinoids (Tab. I). TC and AC are considered well-differentiated neuroendocrine tumors (NETs), clinically of low (corresponding to G1 NET) and intermediate (corresponding to G2 NET) grade, respectively, whereas LCNEC and SCLC are NECs clinically of high grade (traditionally graded as G3 tumors) [28]. This histologically defined classification has strong molecular, clinical and behavioral correlations. TC are low malignant tumors with low mutational burden (TMB) and good prognosis, which are usually cured by surgery alone. AC are intermediately malignant tumors, again with low TMB but more aggressive clinical course and metastatic propensity, which are best treated with surgical resection in early-stage tumors and with inconsistent response to multimodality therapy. In turn, LCNEC and SCLC are clinically aggressive carcinomas with high TMB and dismal prognosis, which can be treated with surgical resection in early stage disease with adjuvant chemo-radiotherapy, but in advanced disease chemo-radiotherapy is the main approach [28]. Since most SCLC present in advanced disease, most are diagnosed on small biopsy or cytology samples. Due to the need for recognizing neuroendocrine morphology and the difficulty in identifying this morphology in cytology or very small biopsy samples, historically LCNEC were usually diagnosed primarily on resection specimens and the diagnosis was seldom made in patients with advanced disease. However, now that larger tissue samples are obtained due to requirments for molecular testing, the diagnosis is able to be established more often in small biopsies and the frequency of LCNEC diagnosis in patients with advanced tumors is destined to increase. Diagnostic criteria for LCNEC in small biopsy samples is challenging, but have been recently proposed [43].

Table 1.

WHO classification of lung neuroendocrine neoplasms and some clinicopathologic correlates.

While necrosis and mitotic count guide classification, cytological criteria split large cell neuroendocrine carcinoma from small cell lung carcinoma. Immunohistochemistry (IHC) for neuroendocrine (NE) markers are defining for large cell neuroendocrine carcinoma only, while contributes to refine diagnosis in the other neoplasms. Typical carcinoid and, especially, small cell carcinoma associate with paraneoplastic syndromes, whereas combined variants of tumors are practically neuroendocrine carcinomas.

Variable	Typical carcinoid	Atypical carcinoid	Large cell neuroendocrine carcinoma	Small cell lung carcinoma
Mitoses per 2 mm 2	0-1	2-10	≥ 11	≥ 11
Necrosis	No	Punctate	Extensive	Extensive to geographic
Cytological details	Variable	Variable	Large cells	Small cells
IHC for NE markers	Contributory to diagnosis	Contributory to diagnosis	Defining for diagnosis	Contributory to diagnosis
Combined variants	No	No	Yes	Yes
Paraneoplastic syndrome association	Uncommon	Rare	Rare	Frequent

THE POSITION OF KI-67 LABELING INDEX INSIDE WHO CLASSIFICATION

Ki-67 antigen is not an essential criteria for diagnosis of lung NENs [28,32], at variance with the homologous lesions arising in the gastroenteropancreatic (GEP) tract where this marker plays a major role in tumor classification [14,44]. However, statistically significant differences can be seen in the Ki-67 labeling indexes among the diverse subtypes of lung NENs, especially with carcinoids [32]. The difficulty in incorporating Ki-67 proliferation rates into diagnostic criteria is largely due to the overlap among the lung NENs (TC vs AC; AC vs LCNEC; LCNEC vs SCLC) [28,32]. This is largely due to the nonlinear relationship existing between Ki-67 index and the key histologic criteria, namely mitoses, necrosis and cell size [32,45]. Secondly, a source of inconsistency is likely owing to the breadth of diagnostic intervals used for classification, especially in AC (2-10 mitoses per 2 mm2 or punctate necrosis) and NECs (over 10 mitoses per 2 mm2 without any upper limit, more marked necrosis and a constellation of tumor cell morphologic features and IHC findings) [28]. For these reasons, there are no consistent thresholds distinguishing TC (G1 NET) from AC (G2 NET) or LCNEC from SCLC. Thirdly, the interobserver variability in the assessment of the Ki-67 labeling index, even while limited to areas of highest staining (methodologically by accounting for the so-called hot spots), and the lack of agreed-upon standards in the relevant quantifications (interpretatively by relying on manual or automated counting) represent further concerns for the systematic use of Ki-67 labeling index as a classifier of lung NENs [32,45]. Fourthly, the relationship between the numerical value of Ki-67 labeling index (expressed as mean, median or continous variable) for the prognostic assessement of tumor subtypes and its lack of consistent demonstration as an independent predictor in multivariable analysis make application of this marker still unclear [46,47]. In other words, Ki-67 has not been shown to improve the existing diagnostic histological criteria, which continue to represent the gold standard and the backbone for classification [28]. In general, it is suggested that a Ki-67 labeling index higher that 5% may correspond to a diagnosis of AC and an index higher than 30% may suggest NECs (either LCNEC or SCLC). Attempts to combine the standard histologic criteria with Ki-67 to refine prognostic categories of NE tumors have been proposed, but the evidence has not been regarded as sufficient to replace the current criteria for separating these tumors, which are primarily based on histologic features alone [48-54].

Since histologic diagnostic criteria for TC and AC were established based on resection specimens, the same criteria should not be applied in the metastatic setting where mitotic rates may be higher than seen in the primary tumor. Similarly, Ki-67 rates may be higher than in the primary tumors and the proliferation rates can vary significantly according to anatomical sites, disease timing and therapy interference [55]. To address this issue, the new 2021 WHO classification has introduced the term “carcinoid tumor, not otherwise specified-(NOS)” for defining carcinoids encountered in the setting of metastases in addition to small biopsies or poorly sampled resection specimens [28]. In these instances, the diagnosis of carcinoid tumor-NOS should be accompanied by reporting the mitoses per 2 mm2, necrosis and Ki-67 labeling index [28].

Despite limitations of the Ki-67 labeling index as a diagnostic criteria for lung NENs, since this is standardly performed in NENs of the GEP tract [56], many medical oncologists ask for this in metastatic carcinoids to better individualize treatments [52,57-61]. Ki-67 labeling index is recommended by the European Society of Neuroendocrine Tumors (ENETS) in the diagnostic evaluation/work-up guidelines of lung NETs [60]. There does not appear to be a meaningful role for the Ki-67 proliferative status in lung NECs to stratify patients for radiotherapy [62], non-conventional chemotherapy in small tumor series [63] or first-line platinum-based chemotherapy for SCLC [64].

HOW TO CORRECTLY APPLY KI-67 LABELING INDEX IN LUNG NENS

An important diagnostic role for Ki-67 in the diagnosis of lung NEN was highlighted by Prof. Rosai and one of us (GP) in a seminal paper [17]. This paper demonstrated the usefulness for Ki-67 in the diagnosis of lung NEN in separating SCLC from carcinoids, particularly in small specimens with crush artifact [17]. This can help avoid the misinterpretation of NETs as NECs, especially SCLC (the most frequent tumor among lung NENs) [17]. Since mitoses and cytological details can be difficult to perceive in small crushed biopsy specimens, the Ki-67 labeling index can be helpful because carcinoids usually show rates that are < 20-30%) while SCLC and LCNEC typically show rates over 50%, often exceeding even 70-80% [17,28,32,65]. So, a NEN with a Ki-67 labeling index less than 20-30% is more likely to be a carcinoid tumor than SCLC or LCNEC. Practical examples of the use of Ki-67 labeling index on biopsy and cytology samples are depicted in Figure 1 A-D and Figure 2 A-D, respectively.

Figure 1.

Representive distribution of Ki-67 labeling index in biopsy of carcinoid with crush artifacts. (A) Beneath the bronchial mucosa that shows squamous metaplasia (curved arrows) there is an extensive infiltrate of crushed tumor cells (arrowheads). A small organoid nest of more preserved tumor cells is present (arrow). (B) Higher power of this area shows how the crush artifact makes it difficult to discern the morphology of the tumor cells and gives an appearance similar to that frequently seen in small cell carcinoma (arrowheads). The focal organoid nest of preserved tumor cells shows moderate eosinophilic cytoplasm and uniform morphology more consistent with carcinoid than small cell carcinoma (arrow adjacent). (C) Chromogranin shows diffuse strong staining confirming the infiltrating cells are from a neuroendocrine tumor. (D) Ki-67 immunostaining shows a very low proliferation index with only one positive tumor cell (arrow) confirming that this is a carcinoid tumor rather than a small cell carcinoma.

Figure 2.

Representive distribution of Ki-67 labeling index in cytology samples of carcinoid and small cell carcinoma. Cytological features of carcinoid (A) may somewhat resemble small cell carcinoma (C), especially when occurring crush artefacts (A, inset). Even in these instances, however, the Ki-67 labeling index is diagnostic by showing very few stained tumor cells in the carcinoid case (B) and numerous elements with nuclear decoration for this markers in small cell carcinoma (D).

DISCUSSION AND CONCLUSIONS

The diagnosis of carcinoid, either TC or AC, and SCLC has consistently been shown to have clinical impact, with profound differences in prognosis and clinical management of patients. Since the criteria for diagnosis of TC and AC were established on resection specimens and the distinction between TC and AC on preoperative biopsies or cytology samples is imprecise [66], the new 2021 WHO classification and ENETS guidelines [28,60] recommend to use the term carcinoid-NOS in small biopsies, cytology and specimens from metastatic samples, by also recording the number of mitoses, the presence or absence of necrosis and the value of Ki-67 (if available) to provide a clearer orientation to clinicians.

The diagnostic and therapeutic decisions for patient management mainly rely on histologic subtyping, but there are at least two areas in the new 5th edition of WHO classification [28] where Ki-67 antigen may play a diagnostic role. First, it is helpful to help avoid overdiagnosing carcinoids as SCLC or LCNEC in small-sized tissue fragments, particularly those with crush artifact. Secondly, another potential role involves the emerging concept of atypical carcinoids with increased mitotic or proliferation rates, which straddle AC and a subset of LCNEC, while preserving morphologic and molecular traits of well-differentiated NETs (see below).

Atypical carcinoids with increased mitotic or proliferation rates

WHO CLASSIFICATION

Rare lung carcinoids, either metastatic or primary, exhibit mitotic counts exceeding the upper threshold for AC (>10 mitoses per 2 mm2) [55,58,67-69]. Although there are no well established cut-off thresholds for lung NENs, AC can have an increased Ki-67 labeling index up to 30%. These tumors are likely to harbor chromatin remodeling-related MEN1, ARID1A, ARID1B, KDM5C mutations [70], while preserving retinoblastoma expression and lacking TP53 inactivation [55]. In keeping with their overall appearance of carcinoids, a new designation of “carcinoid tumors with elevated mitotic counts and/or Ki-67 proliferation rates” has been identified as an emerging concept in the new 5th edition of WHO classification [28]. Such tumors were recognied over 20 years ago in the 1999 WHO classification [71], but due to lack of sufficient data it was proposed that these tumors should be classified as LCNEC. In the current 2021 WHO classification, this recommendation is maintained, with the added comment that the tumor have features of a carcinoid tumor [28,55]. The rarity of these tumors is reflected by the fact that after more than 20 years, there still remain insufficient clinical, pathologic, genetic and epidemiologic data to define a corresponding tumor in the 2021 WHO Classification [71].

HISTOLOGICAL FINDINGS

Representative pictures of AC with elevated proliferation rates primary to the lung are shown in Figure 3 A-D. AC with increased proliferation feature NETs with organoid (trabecular, lobular, palisading, rosettes) to solid patterns of growth, where mitotic count and necrosis exceede what is permitted for AC and/or Ki-67 labeling index is around or over 20-30%. Tumor cells are poligonal to spindled in shape, show eosinophilic cytoplasm, granular to coarser chromatin and variably prominent nucleoli, with easy-to-find mitotic figures. Necrosis is more abundant than AC and may be multifocal. While diagnostic criteria remain to be established, preservation of carcinoid-like morphology with mitotic activity greater than 10 per 2 mm2 or Ki-67 labeling index over 30% appear to be characteristics traits of these tumors [28].

Figure 3.

Carcinoid tumors with elevated mitotic counts and/or Ki-67 proliferation rates. (A, B) This tumor resembles a carcinoid tumor with organoid nests of tumor cells associated with punctate necrosis (arrowheads) and several mitotic figures. The mitotic count for this tumor was 14 per 2 mm2. (C) The tumor cells show morphology of a carcinoid tumor with rosette-like structures (curved arrows) and tumor cells showing finely granular nuclear chromatin with moderate cytoplasm showing an eosinophilic hue. Several mitoses are present (arrows). (D) Immunohistochemistry for Ki-67 shows a proliferation rate of approximately 30%.

IMMUNOHISTOCHEMISTRY AND MOLECULAR FINDINGS

AC with elevated proliferation rates are positive for pan-neuroendocrine markers (chromogranin A, synaptophysin, CD56), somatostatin receptors, but negative for p53 (with normality staining pattern) and with retained retinoblastoma protein [55,67]. The Ki-67 labeling index is increased over 30% [53,55,58,70,72]. The tumor mutation burden per Mb of sequenced DNA is as many as 1.5 [55], somewhat intermediate between carcinoids and NECs [73], and survival is not so poor as classical LCNEC [55,67]. The genetics are more like carcinoids rather than NECs with mutations affecting especially chromatin remodeling genes such as MEN1, ARID1A, ARID1B, KDM5C, while TP53 and RB1 mutations are completely absent [55,70].

In the new 2021 edition of WHO classification on lung tumors, it was recognized that there is an emerging group of primary pulmonary atypical carcinoids with elevated mitotic counts and/or Ki-67 proliferation rates. However, there was insufficient data to define diagnostic criteria for a specific entity and the use of NET G3 was regarded as premature [53,67,69,70,74,75]. The existence of proliferating carcinoids is likely to be an underrecognized phenomenon in thoracic pathology, with instances documented in the lung [55,68,75] and the thymus [76,77]. The emerging recognition of lung AC with increased mitotic or proliferation rates makes up another application of Ki-67 labeling index, which has recently been endorsed by 2021 WHO classification. These AC group with increased proliferation rates is thus an emerging category in the spectrum of lung NENs, that however still require more definite diagnostic criteria and characterization with clinical, genetic and epidemiologic characteristics [68,78,79]. They might occupy a position between NEC and NET arms, but more akin to the latter in terms of biological behavior, molecularly traits and histologic appearance [70,80,81].

Final remarks

The new 5th edition of WHO classification [28] provides important updates regarding lung NENs, with interpretation keys on diagnostic characteristics, the use of Ki-67 staining and the terminology in cytology/biopsy samples, either primary or metastatic. In this scenario, we have outlined the use of Ki-67 labeling index in the theme of differential diagnosis of lung NENs, with a brief revision on the new subsets of AC with increased mitotic or proliferation rates, to help pathologists and clinicians familiarize with these concepts of practical interest in clinical management of lung NEN patients.