Salivary duct carcinoma (SDC) is one of the most aggressive malignancies affecting the salivary glands and histologically resembles high-grade mammary ductal carcinoma.1 Although the majority of SDCs behave in a high-grade fashion with frequent lymph nodal and distant metastases and a dismal prognosis, relatively wide variation in their biological behavior and histologic appearances is recognized.2–4
In addition to surgical resection followed by radiotherapy and conventional chemotherapy, recent advances in molecular-targeted therapy have broadened the therapeutic strategies, with approaches such as trastuzumab targeting the human epidermal growth factor receptor 2 (HER2) or combined androgen blockade targeting androgen receptor.4–8 However, molecular-targeted therapy has not yet been provided worldwide as a standard treatment option, and drug resistance is a perplexing problem that remains to be overcome. Given that accurate risk stratification of patients can help form the basis for treatment decisions, it remains imperative to identify clinical as well as pathologic features that predict the biological behavior of the tumor and the patients’ prognosis.
In breast ductal carcinoma, a histologic grading scheme known as the Nottingham-Bloom-Richardson histologic grading system (Nottingham histologic grade) consisting of 3 histologic factors (nuclear atypia and pleomorphism, mitosis, and tubule formation) is widely used and routinely evaluated in clinical pathology practice.9–11 This histologic grade or risk stratification model is useful for determining the aggressiveness of a tumor and is considered to be a powerful prognosticator, thus playing a key role in guiding clinical decision-making.12 However, a histologic risk stratification model of SDC has not yet been developed, and a limited number of studies have evaluated the prognostic significance of several histologic parameters individually.4,13
In the present study, we evaluated detailed histopathologic indicators in 151 SDC cases associated with the overall survival (OS) and progression-free survival (PFS). Furthermore, based on these results, we attempted to propose a concise and practical histologic risk stratification model of SDC, that can aid in determining the prognosis and the most suitable treatment for each case.
MATERIALS AND METHODS
The present study was approved by the Institutional Ethics Review Board of the ethics committee of our facilities.
The patients of this cohort have been included in previous studies.14,15 This study comprised 151 patients with SDC diagnosed and treated at 7 institutions between 1992 and 2014, except for patients who underwent anti-HER2 or antiandrogen receptor therapy as an initial treatment. All tumors were confirmed to have been diagnosed correctly by 3 expert pathologists (T.N., Y.S., and M.N.) according to the rigorous histomorphologic criteria for SDC.1 Other entities, including the high-grade transformation of various carcinomas, adenocarcinoma, not otherwise specified, high-grade mucoepidermoid carcinoma, and low-grade intraductal carcinoma (previously called as “low-grade SDC”), were carefully eliminated from this study.3
The Evaluation of Histologic Factors
All resected specimens were formalin-fixed, sectioned, and stained with hematoxylin and eosin (H&E). Multistep sections from the whole tumor, including the largest cross-section, were subjected to a histologic review. Each histologic feature was evaluated by 4 pathologists (Y.S., M.N., T.N., and M.U.) independently, and any discrepant results were discussed by an expert panel to achieve consensus. The pathologists were blinded to clinical data at the time of histologic evaluation.
The evaluation criteria of nuclear size and pleomorphism, mitotic count, and tubule formation were primarily based on the Nottingham histologic grade.9–11 In this system, nuclear size and pleomorphism were scored 1 to 3. Small, regular uniform tumor cell nuclei were scored as 1, nuclei with a moderate increase in size were scored as 2, and a score of 3 was given when tumor cells had large nuclei and/or nuclei with marked variation in size and shape. To segregate cases with prominent nuclear pleomorphism, we focused on bizarre tumor cells throughout the tumor, rather than focal, regardless of the absolute nuclear size. These tumor cells had nuclei with marked variation in size and possessed prominent nucleoli. The mitotic count was determined in 10 fields with a ×40 objective lens (HPF). In the Nottingham histologic grade, a score of 1 was given when there were ≤8 mitoses counted, whereas 9 to 17 mitoses were scored as 2, and ≥18 were scored as 3 (field diameter 0.55 mm). In our analysis, high mitotic counts were defined as ≥30 mitoses in 10 HPF (same field diameter).
Lymphatic and vascular invasion was first assessed by H&E staining. An analysis with Elastica van Gieson (EVG) and D2-40 immunohistochemical staining (dilution at 1:100; Dako, Carpinteria, CA) of the representative tumor slide was then performed regardless of the results of the H&E evaluation. Noncomedo necrosis was defined as coagulative tumor necrosis in the invasive component imparting an infarcted appearance in contrast to intraductal comedo necrosis.16 The evaluation of tumor budding was based on the recommendation of the International Tumor Budding Consensus Conference.17 Tumor budding is defined as single cells or clusters of up to 4 cells at the invasive margin. It was counted on H&E staining using a ×20 objective lens and assessed in the highest hotspot at the invasive front. Tumor budding was graded from 1 to 3 according to the number of tumor buds as follows: grade 1, 0 to 4 buds; grade 2, 5 to 9 buds; and grade 3, ≥10 buds. In this study, grade 1 cases and grades 2 and 3 cases were categorized into low tumor budding and high tumor budding categories, respectively. In contrast to tumor budding, a poorly differentiated cluster (PDC) was evaluated as a cancer cell cluster composed of ≥5 cancer cells lacking a gland-like structure, as reported previously.18,19 The counting and grading methods were the same as for tumor budding. The tumor-stroma ratio was assessed using a ×10 objective lens in the most stroma-abundant area. The field with the highest stromal percentage was selected and scored.20,21 A stroma percentage ≤50% was categorized as stroma-low, whereas >50% was categorized as stroma-high.
Univariate and multivariate Cox proportional hazards models and the Kaplan-Meier product-limit method were used to investigate the associations between the histologic features and OS and PFS. The potential confounders in the multivariate analysis included age, sex, primary tumor site, TNM classification, first-line treatment, and histologic origin (ie, de novo or ex pleomorphic adenoma). The strength of an association was determined by the hazard ratio (HR) and the 95% confidence interval (CI). The statistical analyses were performed using the STATA software program (version 13; StataCorp., College Station, TX). All of the tests were 2 sided, and P-values <0.05 were considered to indicate statistical significance.
The clinical characteristics of this study are summarized in Table 1. The mean age was 62.0 years old. The male:female ratio was 127:24, and a strong male predominance was observed. The majority of cases (117 cases, 77.5%) arose in the parotid gland. De novo SDC was noted in 57 cases (37.7%), and preexisting pleomorphic adenoma was recognized in 89 cases (58.9%). The median follow-up period of survivors was 3.4 years (range, 0.04 to 19.0 y). The 3-year OS was 68.5% (95% CI, 60.1%-75.5%), and the 3-year PFS was 34.3% (95% CI, 26.7%-42.1%).
The Nottingham histologic grade (nuclear size and pleomorphism, mitosis, and tubule formation); prominent nuclear pleomorphism; mitosis (≥30/10 HPF); lymphatic, vascular, and perineural invasion; noncomedo necrosis; histologic origin; dominant growth pattern; tumor budding; PDC; and tumor-stromal ratio were evaluated for each tumor. Representative histologic features are shown in Figures 1 and 2.
The distribution and the results of univariate and multivariate analyses of each histologic feature are shown in Table 2 and Supplementary Table S1 (Supplemental Digital Content 1, http://links.lww.com/PAS/A885). On a univariate analysis, prominent nuclear pleomorphism (P=0.004 and 0.007), lymphatic invasion with D2-40 stain (P=0.04 and 0.038), vascular invasion assessed by H&E stain (P=0.003 and <0.001), noncomedo necrosis (P=0.024 and 0.013), dominant invasive growth (P=0.001 and <0.001), and high PDC (P<0.001 and <0.001) were negative prognostic indicators for the OS and PFS. Furthermore, ≥30 mitoses/10 HPF (P=0.007), vascular invasion with EVG stain (P=0.005), perineural invasion (P=0.003), and high tumor budding (P=0.001) were associated with a poor PFS. The loss of tubule formation was associated with a better OS and PFS (P=0.015 and 0.002).
A multivariate analysis showed prominent nuclear pleomorphism (P=0.013 and 0.019), high tumor budding (P=0.011 and <0.001), and high PDC (P<0.001 and <0.001) to be associated with a worse OS and PFS and ≥30 mitoses/10 HPF was correlated with an inferior prognosis for the PFS (P=0.013) (Table 2, Fig. 3). SDC patients with vascular invasion demonstrated poorer PFS with marginal significance (P=0.064).
Histologic Risk Stratification Model
When the SDC cases in this cohort were classified based on the Nottingham histologic grade, grade 1 tumor was only found in 1 case (0.01%), and the HR was not calculable (Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/PAS/A885). To establish a new histologic risk stratification model, we selected 4 histologic features that had an association with a poor OS and/or PFS (P<0.1) in a multivariate analysis: prominent nuclear pleomorphism, mitosis (≥30/10 HPF), vascular invasion, and high PDC. Patients were assigned to 3 risk groups according to the total number of positive factors (among these 4 factors), as follows: low risk, 0 to 1 point (n=35, 23.2%); intermediate risk, 2 to 3 points (n=81, 53.6%); and high risk, 4 points (n=35, 23.2%) (Fig. 4). SDC patients with a higher histologic risk group showed a significantly worse OS and PFS, and the HR progressively increased as histologic risk group increased on univariate and multivariate analyses (Ptrend<0.001) (Table 3, Fig. 5).
The number of studies with histologic assessments of SDC is limited, largely due to the rarity of this entity. Clinically, the tumor size, nodal status, and metastasis have been described as associated with the patient prognosis.15,22–26 We performed a comprehensive histologic analysis of 151 SDC cases to identify histologic features that have a close association with the OS and PFS, independent of Union for International Cancer Control (UICC) TNM factors.27 The items we investigated included classic histologic features related to malignancy, such as nuclear atypia, mitosis, or lymphovascular invasion, and recently established histologic parameters, such as tumor budding and PDC, that have a strong impact on the patient survival in various tumor types.
Histologically, SDC has a striking resemblance to breast ductal carcinoma; however, we felt that SDC exhibited more pronounced nuclear atypia and had more mitoses. Indeed, when classified by the Nottingham histologic grade, only 1 case (0.01%) of SDC was classified as histologic grade 1 (Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/PAS/A885). One previous study applied the Nottingham histologic grade to SDC, and no cases were classified as grade 1.28 These results contrasted with the relatively even distribution of breast carcinoma among grades.9,10 As the Nottingham histologic grade and individual histologic features in this system were not deemed appropriate for SDC, we modified the evaluation criteria of the parameters for breast cancer. We focused only on prominent nuclear pleomorphism throughout the tumor, not the nuclear size or focal change, and increased the mitotic count cutoff value (≥18 in Nottingham histologic grade and ≥30 in our analysis). As tubule formation was associated with a better prognosis in a univariate analysis but not in a multivariate analysis, this histologic feature was not included in the proposed histologic risk stratification model.
In previous reports, the lymphatic, vascular, and perineural invasion was considered to be associated with the prognosis of patient with SDC.29–32 However, whether or not vascular invasion should be separately evaluated from lymphatic invasion and whether or not EVG or D2-40 immunohistochemical staining are necessary have been unclear.13 In the present study, vascular invasion showed a stronger association with the PFS than lymphatic invasion, and this association was also observed in the multivariate analysis, suggesting that vascular invasion should be separately evaluated from lymphatic invasion. As additional EVG staining and D2-40 immunohistochemistry did not predict a poor patient prognosis more precisely than H&E staining, despite the potential for observer-based variation, these additional stains might not necessarily be required to detect lymphatic or vascular invasion (Table 2 and Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/PAS/A885).
In our cohort, 89 SDC cases (58.9%) arose from preexisting pleomorphic adenoma, whereas others were de novo. In line with previous studies, the prognosis of SDC was not influenced by the histologic origin in either a univariate or multivariate analysis.22,30,33,34
SDC with a predominant intraductal component has been considered to have a better prognosis than invasive SDCs.2 We evaluated the amount of intraductal component of each tumor. Thirty-four SDC cases (22.5%) exhibited a dominant intraductal pattern consisting of large ductal structures with a cribriform, papillary, or solid growth (in >50% of the tumor). Although these cases showed a better prognosis than SDC with a dominant invasive growth in univariate analysis, this difference disappeared in multivariate analysis because of the close association with confounders.
Tumor budding and PDC have recently been described as notable histologic features that reflect the type of infiltrative growth at the invasion front and have not been evaluated previously for SDC. Tumor budding is reported to be a promising adverse prognostic indicator in many organs, including the colon, esophagus, breast, skin, stomach, and pancreas.35 PDC is also a poor prognosticator in colorectal cancer.18,19 In the present study, high tumor budding and PDC were strongly related to a poor OS and PFS in univariate and multivariate analyses; as PDC was more prevalent than tumor budding and a high PDC was associated with a higher HR and lower P-value than high tumor budding, we included PDC as an item for determining the histologic risk group. Although many previous studies of the tumor-stroma ratio concluded that high-stromal content was associated with a poor prognosis in other organs,20,21 in our analysis, no significant association was noticed.
The histologic grading or risk stratification model is widely used to predict the prognosis in patients with various malignant tumors. Among salivary gland carcinomas, mucoepidermoid carcinoma has widely recognized histologic grading systems.36 The 2 most popular systems are the point-based manner in which values are assigned for various histologic parameters.37,38 The Brandwein system includes not only features adopted in the AFIP system, such as intracystic component, perineural invasion, necrosis, mitosis, and anaplasia, but also lymphovascular invasion, bony invasion, and invasive front morphology.37,38 In SDC, the application of the Nottingham histologic grade was inappropriate because of the above-mentioned inability to stratify SDC cases, and a histologic risk stratification model specific to this lesion has been warranted.28 We selected the 4 histologic features deemed capable of predicting a poor OS or PFS in a multivariate analysis (prominent nuclear pleomorphism, ≥30 mitoses/10 HPF, vascular invasion, and high PDC), and the number of positive factors was classified 3 different risk groups (Fig. 4). Consequently, this risk stratification model was able to stratify SDC cases evenly into 3 groups of low-risk, intermediate-risk, and high-risk SDC. Using this novel model, risk groups can be assigned based on the findings of a microscopic evaluation with H&E staining alone, making it a useful and practical system that can be applied in any pathology laboratory and which requires no special ancillary testing. Although each histologic feature was a strong prognostic factor, the combination of these 4 histologic features might minimize the intra-observer variation that might arise when relying on a single feature.
The present study was associated with several limitations. Some of the evaluated features, for example, prominent nuclear pleomorphism, could be subjective. The concordance ratio in terms of prominent nuclear pleomorphism was not necessarily high. For example, the agreement rate between 2 pathologists was 82.1% and Cohen κ coefficient was 0.575. To preserve the integrity and prevent interobserver variation, the pathologists discussed discordant cases, and in Figure 1 typical images of prominent nuclear pleomorphism were presented. In addition, to avoid ambiguous usage of the term PDC, we strictly followed the previous criteria and counted the number of PDCs (Figs. 2F–H).18,19 Further validation cohort studies with different datasets and pathologists are warranted to confirm the reproducibility of this risk stratification model. Moreover, as the proposed model is based solely on the histologic features on H&E sections, the relationship with the molecular biomarker profiling classification assessed by the androgen receptor, HER2, and Ki-67 expression status requires further investigation.14
In conclusion, prominent nuclear pleomorphism, ≥30 mitoses/10 HPF, vascular invasion, ≥5 tumor budding, and ≥5 PDCs were strong prognostic predictors of a poor OS or PFS. The histologic risk stratification model based on these factors is a concise and practical method for predicting patient prognosis and providing appropriate therapeutic options.
1. Nagao T, Licitra L, Loening T, et alEl-Naggar AK, Chan JKC, Grandis JR, Takata T, Slootweg PJ. Salivary duct carcinoma
. WHO Classification of Head and Neck Tumours. Lyon, France: International Agency for Research on Cancer; 2017:173–174.
2. Delgado R, Vuitch F, Albores-Saavedra J. Salivary duct carcinoma
. Cancer. 1993;72:1503–1512.
3. Simpson RH. Salivary duct carcinoma
: new developments—morphological variants including pure in situ high grade lesions; proposed molecular classification. Head Neck Pathol. 2013;7(suppl 1):S48–S58.
4. D’Heygere E, Meulemans J, Vander Poorten V. Salivary duct carcinoma
. Curr Opin Otolaryngol Head Neck Surg. 2018;26:142–151.
5. Fushimi C, Tada Y, Takahashi H, et al. A prospective phase II study of combined androgen blockade in patients with androgen receptor-positive metastatic or locally advanced unresectable salivary gland carcinoma. Ann Oncol. 2018;29:979–984.
6. Takahashi H, Tada Y, Saotome T, et al. Phase II trial of trastuzumab and docetaxel in patients with human epidermal growth factor receptor 2-positive salivary duct carcinoma
. J Clin Oncol. 2019;37:125–134.
7. Boon E, van Boxtel W, Buter J, et al. Androgen deprivation therapy for androgen receptor-positive advanced salivary duct carcinoma
: a nationwide case series of 35 patients in The Netherlands. Head Neck. 2018;40:605–613.
8. van Boxtel W, Locati LD, van Engen-van Grunsven ACH, et al. Adjuvant androgen deprivation therapy for poor-risk, androgen receptor-positive salivary duct carcinoma
. Eur J Cancer. 2019;110:62–70.
9. Elston CW, Ellis IO. Pathological prognostic factors in breast cancer. I. The value of histological grade in breast cancer: experience from a large study with long-term follow-up. Histopathology. 1991;19:403–410.
10. Robbins P, Pinder S, de Klerk N, et al. Histological grading of breast carcinomas: a study of interobserver agreement. Hum Pathol. 1995;26:873–879.
11. Pinder SE, Murray S, Ellis IO, et al. The importance of the histologic grade of invasive breast carcinoma and response to chemotherapy. Cancer. 1998;83:1529–1539.
12. Ravdin PM, Siminoff LA, Davis GJ, et al. Computer program to assist in making decisions about adjuvant therapy for women with early breast cancer. J Clin Oncol. 2001;19:980–991.
13. Seethala RR, Altemani A, Ferris RL, et al. Data set for the reporting of carcinomas of the major salivary glands: explanations and recommendations of the guidelines from the International Collaboration on Cancer Reporting. Arch Pathol Lab Med. 2019;143:578–586.
14. Takase S, Kano S, Tada Y, et al. Biomarker immunoprofile in salivary duct carcinomas: clinicopathological and prognostic implications with evaluation of the revised classification. Oncotarget. 2017;8:59023–59035.
15. Otsuka K, Imanishi Y, Tada Y, et al. Clinical outcomes and prognostic factors for salivary duct carcinoma
: a multi-institutional analysis of 141 patients. Ann Surg Oncol. 2016;23:2038–2045.
16. Fisher ER, Palekar AS, Gregorio RM, et al. Pathological findings from the national surgical adjuvant breast project (Protocol No. 4). IV. Significance of tumor necrosis. Hum Pathol. 1978;9:523–530.
17. Lugli A, Kirsch R, Ajioka Y, et al. Recommendations for reporting tumor budding
in colorectal cancer based on the International Tumor Budding
Consensus Conference (ITBCC) 2016. Mod Pathol. 2017;30:1299–1311.
18. Ueno H, Kajiwara Y, Shimazaki H, et al. New criteria for histologic grading of colorectal cancer. Am J Surg Pathol. 2012;36:193–201.
19. Fonseca GM, de Mello ES, Faraj SF, et al. Prognostic significance of poorly differentiated clusters
and tumor budding
in colorectal liver metastases. J Surg Oncol. 2018;117:1364–1375.
20. de Kruijf EM, van Nes JG, van de Velde CJ, et al. Tumor-stroma ratio in the primary tumor is a prognostic factor in early breast cancer patients, especially in triple-negative carcinoma patients. Breast Cancer Res Treat. 2011;125:687–696.
21. Wang K, Ma W, Wang J, et al. Tumor-stroma ratio is an independent predictor for survival in esophageal squamous cell carcinoma. J Thorac Oncol. 2012;7:1457–1461.
22. Lewis JE, McKinney BC, Weiland LH, et al. Salivary duct carcinoma
. Clinicopathologic and immunohistochemical review of 26 cases. Cancer. 1996;77:223–230.
23. Guzzo M, Di Palma S, Grandi C, et al. Salivary duct carcinoma
: clinical characteristics and treatment strategies. Head Neck. 1997;19:126–133.
24. Barnes L, Rao U, Krause J, et al. Salivary duct carcinoma
. Part I. A clinicopathologic evaluation and DNA image analysis of 13 cases with review of the literature. Oral Surg Oral Med Oral Pathol. 1994;78:64–73.
25. Felix A, El-Naggar AK, Press MF, et al. Prognostic significance of biomarkers (c-erbB-2, p53, proliferating cell nuclear antigen, and DNA content) in salivary duct carcinoma
. Hum Pathol. 1996;27:561–566.
26. Jayaprakash V, Merzianu M, Warren GW, et al. Survival rates and prognostic factors for infiltrating salivary duct carcinoma
: analysis of 228 cases from the Surveillance, Epidemiology, and End Results database. Head Neck. 2014;36:694–701.
27. American Joint Committee on Cancer. AJCC Cancer Staging Manual, 7th ed. New York, NY: Springer; 2010.
28. Di Palma S, Simpson RH, Marchio C, et al. Salivary duct carcinomas can be classified into luminal androgen receptor-positive, HER2 and basal-like phenotypes. Histopathology. 2012;61:629–643.
29. Kim JY, Lee S, Cho KJ, et al. Treatment results of post-operative radiotherapy in patients with salivary duct carcinoma
of the major salivary glands. Br J Radiol. 2012;85:e947–e952.
30. Gilbert MR, Sharma A, Schmitt NC, et al. A 20-year review of 75 cases of salivary duct carcinoma
. JAMA Otolaryngol Head Neck Surg. 2016;142:489–495.
31. Han MW, Roh JL, Choi SH, et al. Prognostic factors and outcome analysis of salivary duct carcinoma
. Auris Nasus Larynx. 2015;42:472–477.
32. Roh JL, Lee JI, Choi SH, et al. Prognostic factors and oncologic outcomes of 56 salivary duct carcinoma
patients in a single institution: high rate of systemic failure warrants targeted therapy. Oral Oncol. 2014;50:e64–e66.
33. Lim CM, Hobson C, Kim S, et al. Clinical outcome of patients with carcinoma ex pleomorphic adenoma of the parotid gland: a comparative study from a single tertiary center. Head Neck. 2015;37:543–547.
34. Boon E, Bel M, van Boxtel W, et al. A clinicopathological study and prognostic factor analysis of 177 salivary duct carcinoma
patients from The Netherlands. Int J Cancer. 2018;143:758–766.
35. Berg KB, Schaeffer DF. Tumor budding
as a standardized parameter in gastrointestinal carcinomas: more than just the colon. Mod Pathol. 2018;31:862–872.
36. Seethala RR. Histologic grading and prognostic biomarkers in salivary gland carcinomas. Adv Anat Pathol. 2011;18:29–45.
37. Goode RK, Auclair PL, Ellis GL. Mucoepidermoid carcinoma of the major salivary glands: clinical and histopathologic analysis of 234 cases with evaluation of grading criteria. Cancer. 1998;82:1217–1224.
38. Brandwein MS, Ivanov K, Wallace DI, et al. Mucoepidermoid carcinoma: a clinicopathologic study of 80 patients with special reference to histological grading. Am J Surg Pathol. 2001;25:835–845.