Chemoradiation-Induced Cell Loss in Human Submandibular Glands

The LaryngoscopeLippincott Williams & Wilkins, Inc.© 2005 The American Laryngological,Rhinological and Otological Society, Inc.Chemoradiation-Induced Cell Loss inHuman Submandibular GlandsChristopher A. Sullivan, MD; Robert I. Haddad, MD; Roy B. Tishler, MD, PhD; Anand Mahadevan, MD;Jeffrey F. Krane, MD, PhDObjectives: Chemoradiation-induced xerostomiaaffects approximately 40,000 head and neck cancerpatients annually in the United States. No human histopathologicor immunohistochemical data exist thatcharacterize chemoradiation-related salivary glanddamage. The objective of this study was to describethe histopathologic and immunohistochemical featuresof the non-acute phase of human submandibulargland damage after chemoradiation therapy.Methods: Pathologic materials were retrieved frompatients who had undergone neck dissection afterprotocol-driven chemoradiotherapy for stage IV headand neck cancer at a tertiary head and neck cancerinstitute. Histologic and immunohistochemical analyseswere performed on representative sections ofchemoradiated submandibular glands, and findingswere compared to age- and sex-matched, untreatedcontrol glands. Results: Forty patients were identifiedwho had undergone neck dissection an average of 11weeks after treatment with induction chemotherapyand chemoradiation therapy for non-oral head andneck cancer. In the chemoradiated glands, light microscopicfindings included pronounced acinar cellloss with accompanying ductal metaplasia and ductalproliferation and increased fibrosis, chronic inflammation,nuclear atypia, and cytoplasmic vacuolizationwhen compared with controls. Microvasculardensity was marginally affected by chemoradiation;cytokeratin staining showed preservation of ductalepithelium when compared to controls. Conclusions:Nonacute changes seen in human submandibularglands after chemoradiation therapy are compared toPresented at the Southern Section Meeting of the Triological Society,Miami, FL, January 13–15, 2005.From the Department of Otolaryngology—Head and Neck Surgery,Wake Forest University Baptist Medical Center, Winston-Salem, NorthCarolina (C.A.S.); the Department of Head and Neck Oncology (R.I.H.) andthe Department of Radiation Oncology (R.B.T.), Dana Farber Cancer Institute,the Department of Radiation Oncology, Beth Israel Deaconess MedicalCenter (A.M.); and the Department of Pathology, Brigham and Women’sHospital (J.F.K.), Harvard Medical School, Boston, Massachusetts, U.S.A.Editor’s Note: This Manuscript was accepted for publication February24, 2005.Send Correspondence to Christopher A. Sullivan, MD, Departmentof Otolaryngology—Head and Neck Surgery, Wake Forest UniversityHealth Sciences, Medical Center Boulevard, Winston Salem, NC 27157–0001, U.S.A. E-mail: csulliva@wfubmc.eduDOI: 10.1097/01.MLG.0000163340.90211.87Laryngoscope 115: June 2005958those seen in previously described irradiated animaland human models. Primary dysfunction in humansappears to be related to a reduction in function andnumber of submandibular gland acinar cells. Theductal system appears to be preserved after chemoradiationtherapy. Implications for management of xerostomiaare discussed. Key Words: Chemoradiation,xerostomia, submandibular gland.Laryngoscope, 115:958–964, 2005INTRODUCTIONThe early effects of radiation on salivary glands werefirst characterized by Jean Bergonie in 1911. 1 Since thattime, much progress has been made in understanding theunderlying mechanism of this damage in rats and primates.No human data exist that characterize the histologicor immunohistologic features of chemoradiationtherapy on salivary glands. Chemoradiation therapy forthe treatment of upper aerodigestive tract malignancyinvariably results in xerostomia that affects speech, swallowing,dental health, and quality of life. Treatment strategiesfor this condition have been necessarily based on apreclinical understanding of the pathophysiology of radiationdamage to the salivary glands in animal models.Such strategies have included the use of artificial salivapreparations and sialogogues and, more recently, the prophylacticadministration of cytoprotectants designed toselectively reduce damage to salivary gland cells duringradiation therapy. 2 Sialogogues remain the mainstay oftreatment, although they require frequent administration,are unpredictable in their efficacy, and may haveintolerable side effects. It was the goal of this investigationto describe the histopathologic and immunohistochemicalfeatures of chemoradiation damage in humansubmandibular glands and to compare these results withexisting preclinical animal data and human radiationdata.MATERIALS AND METHODSSubmandibular gland pathology slides and tissue paraffinblocks were obtained from patients who had undergone neckdissection after protocol-driven chemoradiotherapy for stage IVhead and neck cancer at the Dana Farber Cancer Institute andBrigham and Women’s Hospital. Chemotherapy data, radiationdata, surgical data, and age and sex data were obtained fromSullivan et al.: Chemoradiation-Induced Cell Loss

TABLE I.Age, Sex, Primary Type and Site, Tumor Stage, Induction Chemotherapy, and Chemoradiation Data for CRT-SMG Group.Age (y) Sex Primary Type/Site StageInductionChemotherapyConcomitantChemotherapyRadiation(cGy/fractionation)58 M SCCA/Tonsil T3N2bM0 TPF 3 CP 7000/SFX71 M SCCA/Tonsil T4N2cM0 TPF 3 CP 7000/SFX57 M SCCA/BOT T3N3M0 TPF 3 CP 7000/SFX61 M SCCA/Tonsil T4N2cM0 TPF 3 CP 7000/SFX58 M SCCA/Tonsil T2N2bM0 PF 3 CP 7000/SFX63 M SCCA/Tonsil T3N2cM0 PF 3 CP 7000/SFX52 M SCCA/Tonsil T2N3M0 PF 1 T 7230/AFX-CB60 M SCCA/UNK TXN3M0 PF 3 T 7200/AFX-CB57 M SCCA/Tonsil T4N2bM0 PF 3 T 7200/AFX-CB59 F SCCA/Larynx T4N3M0 TPF 3 CP 7000/SFXAFX-CB twice-daily radiation in last 2 weeks of therapy; BOT tongue base; CP carboplatinum, F 5-fluorouracil; P cisplatinum; SFX once-dailyradiation; T taxotere; UNK unknown primary.clinical records. Radiation doses to the submandibular glandswere calculated from original radiation planning port films andtreatment summary data. Control subjects (C) were then selectedfrom a group of non-irradiated, non-chemotherapy-treated patientswho had undergone neck dissection for the treatment oforal cancer. These patients were then age- and sex-matched tothe chemoradiotherapy group (CRT), creating control submandibulargland (C-SMG) and chemoradiation-treated submandibulargland (CRT-SMG) groups for comparison.Immunohistochemistry was performed using the EnvisionPlus/Horseradish Peroxidase system (DakoCytomation Corp.,Carpinteria, CA), and mouse monoclonal antibodies to cytokeratins(clone MNF116; 1:700, DakoCytomation), Ki-67 (cloneMIB-1; 1:200, DakoCytomation), CD31 (clone JC70A; 1:40, Dako-Cytomation), and P63 (clone 4A4; 1:400, LabVision Corp., Fremont,CA). Briefly, paraffin-embedded sections were incubated inhydrogen peroxidase and absolute alcohol for 30 minutes to blockendogenous peroxidase activity. Heat-induced antigen retrievalwas performed in a citrate buffer (pH 6.0) prior to incubationwith the P63 and Ki-67 antibodies using microwave pretreatmentand pressure cooker pretreatment, respectively. Antigen retrievalwas performed by protease digestion for the cytokeratin (10 mindigestion) and CD31 (20 min digestion) antibodies. Tissue sectionswere subsequently incubated with a primary antibody for 40to 60 minutes at 25°C. After TBS rinses, the tissue was incubatedTABLE II.Age, Sex, Primary Type, Primary Site and Staging Information forControl Group (C-SMG).Age Sex Primary Type/Site Stage58 M SCCA/Tongue T1N0M072 M CIS/Buccal T1N0M057 M SCCA/Tongue T2N1M060 M SCCA/Tongue T2N1M058 M SCCA/Tongue T1N0M063 M SCCA/Tongue T3N1M052 M SCCA/Tongue T3N1M060 M SCCA/Tongue T2N1M057 M SCCA/Tongue T2N1M059 F SCCA/Tongue T3N1M0CIS carcinoma in situ; SCCA squamous cell carcinoma.using the Envision Plus secondary antimouse antibody for 30minutes, followed by diaminobenzidine for 10 minutes. Appropriatepositive and negative controls (using normal mouse serum)were stained in parallel. Ki-67 and P63 antibodies were evaluatedfor nuclear staining, and CD31 and cytokeratin antibodies wereassessed for cytoplasmic immunoreactivity.Hematoxylin and eosin–stained sections were analyzedqualitatively and quantitatively for changes in the ratio of serousto mucinous cells (S:M), degree of inflammation (CI), fibrosis (F)and gain or loss of acinar cells (A), intercalated duct cells (ICD),striated duct cells (SD), excretory duct cells (ED), and adiposecells (Ad) in the CRT-SMG and the C-SMG groups. Quantitativeestimates of the percentage of epithelial cells staining for cytokeratinsand P63 were performed; cell proliferation indices (Ki-67) and microvessel density (CD31) were averaged from counts offive random high-power fields (field size 0.173 mm 2 ; 400 X) inCRT-SMG and C-SMG tissue. The data from the C-SMG andCRT-SMG groups were then compared to each other and to historicaldata in animal and human studies. All patient data acquisitionwas overseen by the Brigham and Women’s InstitutionalReview Board and was HIPPA compliant.TABLE III.Summary of Procedures Used to Obtain Submandibular Tissue,Number of Days between Completion of Chemoradiation Therapyand Histopathologic Analysis of Submandibular Glands, andRadiation Dose to the Submandibular Glands.Procedure Days Out from CRT XRT Dose to SMG (cGy)R MRND 77 7000R MRND 66 7000L RND 72 5600L RND 87 7000R RND 43 5670L RND 107 5580L MRND 123 7200R RND 74 7200R MRND 66 5000L MRND 42 5580CRT chemoradiation therapy; L left; MRND modified radicalneck dissection; R right; RND radical neck dissection; SMG submandibulargland; XRT radiation therapy.Laryngoscope 115: June 2005Sullivan et al.: Chemoradiation-Induced Cell Loss959

TABLE IV.Control Submandibular Gland Immunostaining and Histologic Data.Cytokeratin (%) Ki-67 (cells/HPF) P63 (%) CD31 (vessels/HPF) S:M F CI30 4 10 36 95:5 0 075 2 15 31 90:10 30 2 10 19 80:20 0 015 1 10 25 75:25 0 20 1 10 32 100:0 0 030 2 10 30 80:20 0 060 4 10 34 60:40 0 045 20 25 46 60:40 020 5 12 32 80:20 0 20 6 10 30 90:10 0 0CD31 number of vessels per high-power field staining for CD31; CI chronic inflammation, scale of 0 to ; Cytokeratin % of epithelial cells stainingfor cytokeratin; F fibrosis, scale of 0 to ; Ki-67 number of cells per high power field staining for Ki-67; P63 % of cells staining for P63; S:M ratioof serous to mucinous cells.RESULTSForty patients were identified who had undergoneprotocol-driven induction chemotherapy followed by chemoradiationtherapy for stage IV non-oral squamous cellcarcinoma. From this group, 11 patients were identifiedwhose radiation port films were available for review. Atotal of 12 submandibular glands from 11 patients wereidentified for histopathologic and immunohistochemicalanalysis. Of these, 10 were able to be age- and sexmatchedto control subjects. Age, sex, primary tumor site,TNM stage, chemotherapy, and radiation data are summarizedin Table I and Table II. There were nine male andone female subjects. Average age was 60 years. All patientshad stage IV disease of the tongue base, tonsil, orlarynx. Patients received cisplatin, 5-fluorouracil, andtaxotere (TPF) or cisplatin and 5-fluorouracil (PF)–basedchemotherapy. Submandibular gland specimens were distributedequally between left and right sides. Radiationdoses to the submandibular glands varied between 5000cGy and 7200 cGy with an average dose of 6283 cGy. Thetime from the end of chemoradiation therapy to removal ofthe study glands (CRT-SMG) during neck dissectionranged between 42 and 123 days, with a median of 73 days(10 wk) and an average of 76 days (11 wk) (Table III).The C-SMG group specimens showed normalappearingsubmandibular tissue in all cases (Table IV,Fig. 1) In the CRT-SMG group histopathologic findingsincluded severe loss of both serous and mucinous acinarcells and secretory granules. There was moderate to severechronic inflammation and fibrosis in all CRT-SMGspecimens. (Fig. 2, Table V). In CRT-SMG specimens,cytokeratin (PAN-K) immunostaining of epithelial cellswas increased by a factor of 2.2 corresponding to themorphological appearance of acinar cell loss with accompanyingductal cell metaplasia and ductal proliferation.Cell proliferation as assessed by Ki-67 immunostainingwas increased by a factor of 3.6. P63 staining was increasedby a factor of 2.9 and endothelial cell (CD31)staining was increased by a factor of 1.4. (Table V, VI;Figs. 3, 4, 5, 6).Fig. 1. Representative, normal human submandibular gland fromthe C-SMG group. Note the abundant serous acini with zymogengranules, normal mucinous acinar cells, and ductal elements.Fig. 2. Chemoradiated submandibular gland. Note severe loss ofserous and mucinous acinar cells, ductal metaplasia and proliferation,periductal fibrosis, and lymphoplasmacytic infiltration.Laryngoscope 115: June 2005960Sullivan et al.: Chemoradiation-Induced Cell Loss

TABLE V.Chemoradiation-Treated Submandibular Gland Immunostaining and Histologic Data.Cytokeratin (%) Ki-67 (cells/HPF) P63 (%) CD31 (vessels/HPF) S:M F CI100 19 50 51 0:0 90 19 25 26 S:0 90 10 30 0 0:0 , AI100 4 50 22 0:0 70 35 40 43 0:100 95 14 40 22 10:90 60 21 30 42 S:0 70 18 25 46 0:100 100 20 25 46 S:0 100 19 30 43 0:0 , AIAI focal acute inflammation; CD31 number of vessels per high-power field staining for CD31; CI chronic inflammation, scale of 0 to ;Cytokeratin % of epithelial cells staining for cytokeratin; F fibrosis, scale of 0 to ; Ki-67 number of epithelial cells per high-power field staining for MIB-1;P63 % of cells staining for P63; S:0 rare serous cells; S:M ratio of serous to mucinous cells.DISCUSSIONThe existing experimental data on salivary gland radiosensitivityand the theoretical mechanism by which salivarytissue damage occurs during head and neck radiationare well described in animal and primate models. It is generallyagreed that radiation causes lipid peroxidation that iscatalyzed by heavy metal ions present in the granules ofserous cells. This process leads to membrane disruption andrelease of cellular enzymes, which then cause autolysis andcell death. It is hypothesized that this damage may alsoinclude reproductive death of acinar progenitor cells. 3–5 Inhumans, the major salivary glands typically lie within thehead and neck radiation fields, and the amount of glandulardamage is believed to be dependent on radiation dose, fractionation,latency of treatment, the premorbid functionalstate of the glands, and the ratio of serous to mucinous cellswithin an individual gland. 5 Clinical studies have shownboth qualitative and quantitative changes in the saliva ofpatients undergoing head and neck radiation, but data onthe accompanying histopathologic changes is lacking. 6,7Data on chemoradiation-induced histopathologic changes ineither human or animal models is nonexistent. We chose toacquire submandibular tissue from patients who had previouslyundergone protocol-driven treatment because this populationwas essentially homogeneous relative to primary sitetreated and was easily accessible through our protocol database,and because retrospective data acquisition would notrequire any alteration of the cancer treatment plan. Werecognize that prospective data collection could allow forbetter control of variables such as primary site treated, thetiming of concomitant chemotherapy, the agents used, thedose and timing of radiation and control of the time periodbetween the last dose of radiation and the removal of submandibulargland specimens. It is conceivable that patientscould be enrolled in a protocol that would require neck dissectionat a defined number of weeks after chemoradiationtherapy; however, patients show different rates of healingand variable nutritional statuses after chemoradiotherapyand not all patients will be ready to undergo surgery at thetime defined by such a protocol. The logistics of patienttravel and operating room scheduling would make protocolmanagement problematic and could potentially impact cancertreatment outcome unfavorably. It was our objective toidentify histopathologic and immunohistochemical changesthat are seen after chemoradiation therapy for non-oral cancersof the head and neck, to compare these finding to thoseseen in radiation-induced models, and to discuss the impactof these findings on existing and potential treatment optionsfor xerostomia.Only one human study exists that classifies thehistopathologic and immunohistochemical changes tothe salivary glands after therapeutic radiation for headand neck cancer. 4 In this study, radiation damage wasanalyzed using quantitative estimates of the degree ofTABLE VI.Average Values for Quantitative Estimates of Immunostaining Patterns and Histopathologic Findings in Control vs. ChemoradiatedSubmandibular Glands.*Cytokeratin (%) Ki-67 (cells/HPF) P63 (%) CD31 (vessels/HPF) S:M F CIC-SMG 40 5 12 30 81:19 CRT-SMG 88 18 35 42 1:1 RIF 2.2 3.6 2.9 1.4 — — —*Note high relative increase factor (RIF) across all immunostained specimens in the chemoradiation treated group.CD31 number of vessels per high-power field staining for CD31; CI chronic inflammation, scale of 0 to ; CRT-SMG chemoradiation-treatedsubmandibular gland group; C-SMG control submandibular gland group; F fibrosis; Ki-67 number of epithelial cells per high-power field staining for Ki-67;PAN-K % of epithelial cells staining for cytokeratin; P63 % of cells staining for P63; RIF relative increase factor of immunostained cells; S:M ratio of serousto mucinous cells.Laryngoscope 115: June 2005Sullivan et al.: Chemoradiation-Induced Cell Loss961

Fig. 3. Cytokeratin immunostaining of CRT-SMG (A) and C-SMG(B). Note intense cytokeratin staining of the treated gland with arelative decrease in acinar tissue when compared to the controlspecimen.Fig. 4. Ki-67 proliferation immunostaining of CRT-SMG (A) andC-SMG (B). Note the relative increase in Ki-67 staining cells in theCRT-treated specimen.cellular and tissue damage; the degree of damage wasthen classified by severity into three stages. These datadid not include the radiation doses or latency of radiationtherapy. Our histopathologic data correlate wellwith this staging system and in all cases would be bestdescribed as stage III (most severe), the chief features ofwhicharehigh-gradeparenchymalreductionwithlymphocyticinfiltrate, ductal proliferation, periductal fibrosis,andvascularsclerosis.Withtheadditionofweeklychemotherapyto radiation doses between 5000 cGy and 7200cGy, we observed severe and equal loss of both serousand mucinous acinar cells and near complete loss ofsecretory granules between 42 and 123 days posttreatment.These finding are contrasted to rodent andprimate data in which zymogen granule–laden serousacinar cells are considered to be more sensitive to radiationthan mucinous cells and ductal structures. 3,8 Ourdata showed consistent severe loss of both cell types across awide range of radiation doses with complete reversal of theserous to mucinous cell ratio, suggesting that the addition ofchemotherapy may have led to increased acinar cell loss ofboth types when compared with radiation alone. This likelyLaryngoscope 115: June 2005962reflects the known tissue sensitizing effect of chemotherapythat is characterized by greater cell loss in general duringradiation therapy.Immunohistochemical markers for cytokeratin, P63,Ki-67, and CD31 were employed in this study to analyzecellular changes in a semiquantitative manner. Cytokeratinis an intermediate filament protein that makes up acritical component of intracellular mechanical support forepithelial cells. Increased cytokeratin expression in ratsubmandibular cells that have received greater than 40Gy correlated with morphologic changes that were believedto represent irreversible damage to the salivaryglands 4 to 6 months after radiation. 9 In our CRT-SMGspecimens, cytokeratin staining was increased by a factorof 2.2 over control specimens and was seen primarily inassociation with ductal epithelium. Morphologically, thesefindings corresponded to a loss of acinar cells with accompanyingductal proliferation. These alterations would beassociated with a potentially irreversible acinar cell lossand decreased function of the gland.The increased cytokeratin staining seen in CRT-SMGspecimens was accompanied by increased staining for Ki-Sullivan et al.: Chemoradiation-Induced Cell Loss

Fig. 5. P63 immunostaining of CRT-SMG (A) and C-SMG (B). Notethe increase in P63-positive cells in the CRT-treated gland.Fig. 6. CD31 immunostaining of CRT-SMG (A) and C-SMG (B). Noteslightly more extensive staining of endothelial-lined luminal structuresin CRT-treated specimens.Laryngoscope 115: June 200567. Ki-67 is a proliferation-associated antigen that hasbeen used as a marker for sites of cellular proliferation innumerous tissues, including salivary gland. 10 Topographiclocalization and quantification of Ki-67 associatedcellular proliferation in normal human salivary glandshas demonstrated baseline proliferative activity in basal,intercalated, and acinar cells and lower proliferative ratesin myoepithelial and oxyphilic cells. The exact mechanismof physiologic ductal regeneration and metaplasia has notbeen established, but regeneration from an uncommittedstem cell population has long been favored. 11 Based onthese data, Ki-67 staining in our study may representeither normal or pathologic attempts by all or some celltypes to regenerate normal epithelium from undamagedprogenitor cells after chemoradiation therapy.It is possible that a certain number of progenitor cellsremain after radiation or chemoradiation therapy. P63has been shown to stain a population of progenitor stemcells in squamous epithelium and has been shown to be amarker of myoepithelial differentiation in salivary glandand breast tissue. 12–14The combined Ki-67 and P63 immunostaining resultsin our study may suggest both myoepithelial and progenitorcell attempts at regeneration and proliferation. Furtherstudy in human and animal models is needed to provethis concept.It is acinar cell loss and subsequent failure to regenerateacinar structures after radiation or chemoradiationtherapy that ultimately results in the clinical condition ofxerostomia. The additional loss of blood flow to the fibrotic,end-stage gland further limits the stimulated flowof saliva from any remaining acini. Preclinical and clinicalstrategies for preventing xerostomia have been designedto limit cell loss and have used agents that degranulateserous cells, stimulate progenitor cell growth, chelateheavy metals, or scavenge free radicals. Modulated radiationstrategies have attempted to shield the salivaryglands from radiation and thus prevent cell loss. 4 Thedata presented in our study show preservation of theductal system after chemoradiation therapy and thereforea novel approach to xerostomia could be to regenerateacinar cells rather than risk undertreating disease to preserveacinar cells. The CD31 immunostaining data in ourstudy showed that endothelial cell staining was actuallymodestly increased after therapy, suggesting that theremay be adequate vascularity in the salivary bed to supportSullivan et al.: Chemoradiation-Induced Cell Loss963