Histopathology 2002, 41, 152–157
Tumour-associated tissue eosinophilia as a prognostic factor in oral squamous cell carcinomas R G Dorta, G Landman,1 L P Kowalski,2 J R P Lauris,3 M R D O Latorre4 & D T Oliveira Department of Oral Medicine, Area of Pathology, Bauru Dentistry School, University of Sa˜o Paulo, Bauru, 1Department of Pathology and 2Department of Head and Neck Surgery and Otorhinolaryngology, Cancer Hospital, Sa˜o Paulo, and 3 Department of Paediatric Dentistry, Orthodontics and Public Health, Bauru Dentistry School, and 4Department of Epidemiology, School of Public Health, University of Sa˜o Paulo, Sa˜o Paulo, Brazil Date of submission 11 December 2001 Accepted for publication 10 April 2002
Dorta R G, Landman G, Kowalski L P, Lauris J R P, Latorre M R D O & Oliveira D T (2002) Histopathology 41, 152–157
Tumour-associated tissue eosinophilia as a prognostic factor in oral squamous cell carcinomas Aims: Tumour-associated tissue eosinophilia has been described in many sites, including head and neck. The mechanism of eosinophil recruitment and its role in tumours has not yet been defined, and its presence has been related to a favourable as well as unfavourable prognosis. The aim of this study was to evaluate the influence of tumour-associated tissue eosinophilia on the prognosis of 125 oral squamous cell carcinoma patients. Methods and results: The number of eosinophils was obtained by morphometric analysis and ranged from 0 to 392 per mm2. Tumour-associated tissue eosinophilia
was classified according to intensity as mild, moderate, or intense and correlated statistically to the intensity of the mononuclear inflammatory infiltrate as well as to the location of the eosinophilic inflammatory infiltrate. The multivariate analysis demonstrated that intense tumour-associated tissue eosinophilia is an independent favourable prognostic factor for oral squamous cell carcinomas. Conclusion: These findings suggest an anti-tumoral role of eosinophils not as yet well understood that should be better investigated.
Keywords: eosinophilia, eosinophil, squamous cell carcinoma, oral cancer, prognosis
Introduction Tumour-associated tissue eosinophilia is characterized by the presence of eosinophils as a component of peri- and intratumoral inflammatory infiltrate. It has been reported not only in the mouth,1–5 but also in many sites such as oesophagus,6 larynx,1,2,7–10 pharynx,1,2,7,11,12 skin,13 breast,5,14 lung,15 gastrointestinal,4,5,16–19 and genitourinary tracts.4,20–22 Eosinophils are cells derived from the bone marrow23 characterized by the presence of specific granules containing cationic proteins such as the major basic protein, eosinophil cationic protein, eosinophil-derived Address for correspondence: Dr Denise Tostes Oliveira, Av. Getu´lio Vargas 4–82, apto. 301B, CEP 17017-000, Bauru, Sa˜o Paulo, Brazil. e-mail: [email protected]
2002 Blackwell Science Limited.
neurotoxin and the eosinophil peroxidase,23,24 which are strongly stained by eosin.23 Other inflammatory mediators are also synthesized and released by eosinophils, such as granulocyte-macrophage colony-stimulating factor (GM-CSF), interleukin-3 (IL-3), IL-5,24–26 tumour necrosis factor-alpha (TNF-a),23,24,27 transforming growth factor-alpha (TGF-a), and TGF-b.23,24 Some of these factors are related to cell lysis, including tumour cells. Although the exact role of eosinophils in tumours is not yet defined, it has been related to a good,1,2,5,8,10,16–19 to a poor prognosis,3,21,22 or even having no influence on patients’ outcome.9,11,12 One possible explanation for this controversy is the fact that the definition and measurement of tumour-associated tissue eosinophilia is not uniform among authors, precluding a comparison of the results obtained from different studies.
In order to contribute further to the establishment of a role in tumour-associated tissue eosinophilia, we performed this study aiming to: (i) quantify, by morphometric analysis, tumour-associated tissue eosinophilia in oral squamous cell carcinomas; (ii) verify the association between tumour-associated tissue eosinophilia, clinical and microscopic neoplastic characteristics, as well as tumour tissue (acute or chronic) inflammatory response; (iii) verify the influence of tumour-associated tissue eosinophilia on the prognosis of oral squamous cell carcinoma patients.
Patients and methods This study was based on the analysis of 125 patients previously studied by Kowalski et al.28 All the patients were submitted to surgical treatment for primary oral squamous cell carcinoma at the Head and Neck Surgery and Otorhinolaryngology Department of the Cancer Hospital A.C. Camargo, Fundac¸a˜o Antonio Prudente, in Sa˜o Paulo, Brazil, from March 1970 to December 1992. The inclusion criteria were: (i) tumour located in the oral tongue, floor of the mouth, retromolar area and inferior gingiva confirmed by biopsy; (ii) patients not submitted to previous treatment; (iii) patients submitted to surgery as the initial treatment followed or not by radiotherapy; (iv) TNM stage II and III tumours; (v) complete clinical follow-up; and (vi) tumour tissue available for microscopic analysis (glass slides and ⁄ or paraffin blocks). Exclusion criteria were: (i) patients to whom other than surgery was indicated; (ii) other simultaneous primary tumours; (iii) previous chemotherapy; (iv) patients who refused surgical treatment; and (v) tumours with extensive ulceration and ⁄ or necrosis. Clinical data of the patients were collected from the hospital records and included age, gender, ethnic group, tobacco and alcohol consumption, previous oral lesions, blood eosinophilia (values >5%), tumour location, extension and infiltration pattern (ulceroinfiltrative or exophytic), T and N stages, treatment, and clinical follow-up (recurrence, occurrence of a second primary tumour and death). Surgically resected specimens were obtained from the Department of Pathology files, and the microscopic features were evaluated qualitatively by three observers (R.G.D., G.L., and D.T.O.) from the analysis of one 3-lm section of each case, stained routinely with haematoxylin and eosin. Tumour morphology, infiltration of adjacent structures, inflammatory infiltrate distribution and location were reported. Tumour-associated tissue eosinophilia was evaluated quantitatively by using a 25-point ocular graticle 2002 Blackwell Science Ltd, Histopathology, 41, 152–157.
under · 800 magnification. All the specimens were examined through the entire depth, including malignant cells and tumour stroma, covering a total area of 1.32 mm2 per tumour, corresponding to 75 random microscopic fields. The number of observed eosinophils divided by the total area analysed was calculated in each case. Three degrees of tumour-associated tissue eosinophilia intensity were established: absent ⁄ mild, moderate or intense based on the tertiles of the mean number of eosinophil per mm2 obtained in the 125 analysed cases. v2 test was used to analyse the relation between clinical and microscopic variables with tumour-associated tissue eosinophilia. Disease-free survival and overall survival were calculated by the Kaplan–Meier product-limit actuarial method and the comparison of the survival curves was performed using log rank test. Prognostic factors were evaluated in multivariate analysis by the Cox proportional hazard regression model. Patients were followed from the date of surgery to the last objective follow-up information or death.
Results The analysis of 125 patients with oral squamous cell carcinoma revealed a male predominance (84%) over females (16%). Age varied from 30 to 95 years (mean ¼ 58 years). The investigation of oral lesions previous to the diagnosis of oral squamous cell carcinomas was rarely documented (4.8%). Blood eosinophilia was identified in 34.4% of cases. Clinically, lesions were ulcero-infiltrative and exophytic in 32% and 67.2% of cases, respectively. Based on the UICC (International Union Against Cancer) classification of oral cavity carcinomas, tumours were classified as T1 (0.8%), T2 (65.6%), T3 (33.6%), N0 (52.8%) and N1 (47.2%). Recurrence was observed in 46.4% of cases, occurring in the primary tumour site (51.7%), contralateral neck (20.7%), ipsilateral neck (12.1%) and distant sites (5.2%). In 10.3% of the patients, recurrence was seen in more than one location. A second primary tumour occurred in 18.4% of patients, the mouth and pharynx (25%), oesophagus (20%) and larynx (15%) being the most common locations. Details concerning clinical features, treatment and patients’ outcome are summarized in Table 1. The microscopic analysis revealed a diffuse distribution of the inflammatory infiltrate in 94.4% of cases, with an intense mononuclear infiltrate in 67.2% and a mild neutrophilic infiltrate in 69.6%. Eosinophils were present not only in tumour stroma, but within malignant islands as well (45.6%). Moderate
R G Dorta et al.
Oral tongue Floor of the mouth Retromolar area Inferior gingiva
78 28 11 8
62.4 22.4 8.8 6.4
Yes No Unknown
105 8 12
84 6.4 9.6
Yes No Unknown
84 28 13
67.2 22.4 10.4
Ipsilateral Bilateral None
101 21 2
80.8 16.8 2.4
Alive, NED Alive, RD Dead, PO Dead, primary disease Dead, NRD
30 1 8 51 35
24 0.8 6.4 40.8 28
Table 1. Squamous cell carcinoma—clinical findings, treatment and patients’ outcome
NED, no evidence of disease; RD, recurrent disease; PO, post-operative; NRD, not related to disease.
hyperchromatism and pleomorphism were identified in 47.2% of the sample, and few and numerous keratin pearls were equally found (41.6% each). A small percentage of cases (16%) presented high mitotic rate. Perineural invasion (70.4%), vascular embolization (63.2%), muscular (84.8%) and osseous infiltration (7.2%) were also seen. From the 122 patients submitted to ipsilateral neck dissection, the microscopic analysis demonstrated the absence of regional metastasis in 52.5%. From those 21 submitted to contralateral neck dissection, 71.4% did not have regional metastasis. The morphometric analysis revealed a mean of 80.4 eosinophils per mm2, varying from a minimum of 0 to a
maximum of 392 eosinophils per mm2. Tumourassociated tissue eosinophilia was classified according to its intensity based on the tertiles of the number of eosinophils obtained per area. Thus, tumour-associated tissue eosinophilia was considered absent ⁄ mild when a mean of 0–26 eosinophils ⁄ mm2 were observed, as moderate with the presence of 27–83 eosinophils ⁄ mm2, and intense with 84 eosinophils ⁄ mm2 or higher values. From all the clinical and microscopic characteristics studied, mononuclear inflammatory infiltrate intensity and eosinophilic infiltrate location were the only variables statistically associated with tumour-associated tissue eosinophilia (P < 0.001). It revealed a trend 2002 Blackwell Science Ltd, Histopathology, 41, 152–157.
toward a more intense tumour-associated tissue eosinophilia as the mononuclear inflammatory infiltrate increased, and a tendency to eosinophil accumulation exclusively in tumoral stroma when tumour-associated tissue eosinophilia was absent ⁄ mild ⁄ moderate, and an accumulation both in neoplastic clusters and tumour stroma in cases in which tumour-associated tissue eosinophilia was intense (P < 0.05). Survival analysis showed that disease-free survival ranged from 0 to 287.4 months (mean ¼ 137.9 months) and overall survival ranged from 0 to 287.4 months (mean ¼ 88.2 months). Survival rate curves according to Kaplan–Meier method demonstrated a 5-year disease-free cumulative survival of 32%, 44% and 72% for absent ⁄ mild, moderate and intense tissue eosinophilia, respectively (P ¼ 0.024) (Figure 1). The overall cumulative survival for the same period was 27%, 37%, 63% for absent ⁄ mild, moderate and intense tissue eosinophilia, respectively (P ¼ 0.015) (Figure 2). It means that those patients with higher numbers of tumour-associated eosinophils had a better prognosis than patients with intermediate or low counts. Multivariate analysis also revealed that tumour-associated tissue eosinophilia is an independent prognostic factor when matched by age, gender, alcohol or tobacco consumption, tumour site, T and N clinical stage and vascular embolization.
Discussion Although literature demonstrates a tendency to consider tumour-associated tissue eosinophilia as a favourable prognostic factor in head and neck squamous
Figure 1. Prognostic influence of tumour-associated tissue eosinophilia in oral squamous cell carcinomas for disease-free survival (P ¼ 0.024). 2002 Blackwell Science Ltd, Histopathology, 41, 152–157.
Figure 2. Prognostic influence of tumour-associated tissue eosinophilia in oral squamous cell carcinomas for overall disease (P ¼ 0.015).
cell carcinomas,1,2,6–8,10 tumour-associated tissue eosinophilia has also been related to a poorer prognosis3 or even with no influence on patients’ outcome,5,9,12 reflecting that this issue is still a matter of controversy. The definition of tumour-associated tissue eosinophilia itself is debatable. Leighton11 characterized tumour-associated tissue eosinophilia as the tumoral infiltration by eosinophils not related to the presence of necrosis and ⁄ or ulceration. Although authors such as Lowe4 and Looi12 shared this same definition of tumour-associated tissue eosinophilia, this exclusion criterion has not been applied to the majority of studies.1–3,5–10,29 In excluding extensive areas of ulceration and ⁄ or necrosis we aimed to avoid inadequate microscopic analysis of tumoral characteristics. Furthermore, we detected that these areas were not necessarily infiltrated by eosinophils, an opinion shared by Looi.12 Considering specifically the degree of eosinophilic infiltration, parameters available in the literature for intense tumour-associated tissue eosinophilia in head and neck tumours range from >10 up to 100 eosinophils per high-power field.4,9,11 Beyond that, omission of relevant data such as the number of microscopic fields or the total area analysed1,3–5,9–12 allows no comparison among results obtained from different authors. It also seems that biopsy-based studies3,7,9–12 do not reflect precisely the occurrence of tumour-associated tissue eosinophilia, since in our sample, evaluating surgically resected specimens, eosinophils predominated on the tumour invasion front, and biopsy specimens, which are generally small sized, represented only
R G Dorta et al.
the superficial portions of tumours. In the same way, counting exclusively stromal infiltrating cells6 may also underestimate tissue eosinophilia, since 45.6% of our sample had both tumoral stroma and neoplastic eosinophil infiltration, being intimately associated with tumour cells. For the above reasons, we decided to perform an objective and reproducible evaluation of tumourassociated tissue eosinophilia in stage II and III oral squamous cell carcinoma by means of morphometric analysis. Since TNM is a well-established prognostic factor, the option for these intermediate clinical stages lessens the determinant influence of initial and advanced clinical stages. In addition, the objective methodology used allows minimizing the subjectivity of the analysis, prompting the comparison of our results with future studies, an essential step for the establishment of a definite role of eosinophils in squamous cell carcinomas. To our knowledge, this is the first morphometric study that attempts to relate tumour-associated tissue eosinophilia in oral squamous cell carcinoma to prognosis. In this study, we found a striking variation in tumour eosinophil infiltration from 0 to 392 eosinophils ⁄ mm2 (mean 80.4 eosinophils ⁄ mm2). Our classification of tumour-associated tissue eosinophilia based on the tertiles of the number of eosinophils per mm2 present in each of the 125 cases studied allowed us to compare discrete (0–26 eosinophils ⁄ mm2) with intense (‡ 84 eosinophils ⁄ mm2) degrees of tumourassociated tissue eosinophilia regarding prognostic influence. As a result, we demonstrated the influence of high-grade tumour-associated tissue eosinophilia toward a better prognosis of oral squamous cell carcinoma patients. Although further analyses are required to confirm these results, many mechanisms by which eosinophils could play a protective role against tumour progression have been contemplated in the literature. It varies from the capacity of eosinophils to perform antibodydependent cell-mediated cytotoxicity to its capacity of synthesize and release proteins such as major basic protein,4,7,24,30 eosinophil cationic protein,30 and eosinophil peroxidase,24,25 known to have cytotoxic properties, as well as cytokines such as IL-1, TNF-a, IL-3, GM-CSF, TGF-b,23,24,26,27 that could contribute to tumour cytotoxicity directly acting against tumour cells or indirectly, through the autocrine activation of eosinophils or even stimulating additional effector cells.30 However, which cells are responsible for the secretion of eosinophilotactic factors in tumours is still unknown, making this issue even more intriguing.
From this point of view, more detailed research on the influence of mononuclear infiltration on tumour biological behaviour is necessary, since our study demonstrated a trend toward a more intense tumourassociated tissue eosinophilia as the mononuclear inflammatory infiltrate increases (P < 0.001). This finding raises the possibility that tumour-associated tissue eosinophilia may be related to the release of lymphocyte-derived eosinophilotactic cytokines, such as lymphocyte chemoattractant factor, IL-2, IL-3, IL-5 and RANTES,26 or even IL-4, that seem to regulate the expression of VCAM-1, increasing eosinophil transmigration through endothelium.23,25,26,30 In conclusion, although our study has shown that high rates of tumour-associated tissue eosinophilia could represent a favourable prognostic factor in clinical TNM stage II and III oral squamous cell carcinomas from the floor of the mouth, oral tongue, retromolar area, and inferior gingiva, further clinical studies and knowledge about the cellular and molecular interactions of eosinophils in tumour sites are necessary and may contribute in the future to treatment and establishment of prognosis for these neoplasms.
Acknowledgements This study was funded by CAPES (Coordenac¸a˜o de Aperfeic¸oamento de Pessoal de Nı´vel Superior), Brazil. The authors thank Fundac¸a˜o Antonio Prudente.
References 1. Goldsmith MM, Cresson DH, Askin FB. The prognostic significance of stromal eosinophilia in head and neck cancer. Otolaryngol. Head Neck Surg. 1987; 96; 319–324. 2. Goldsmith MM, Belchis DA, Cresson DH, Merritt WD, Askin FB. The importance of eosinophil in head and neck cancer. Otolaryngol. Head Neck Surg. 1992; 106; 27–33. 3. Horiuchi K, Mishima K, Ohsawa M, Sugimura M, Aozasa K. Prognostic factors for well-differentiated squamous cell carcinoma in the oral cavity with emphasis on immunohistochemical evaluation. J. Surg. Oncol. 1993; 53; 92–96. 4. Lowe D, Fletcher CDM. Eosinophilia in squamous cell carcinoma of the oral cavity, external genitalia and anus—clinical correlations. Histopathology 1984; 8; 627–632. 5. Pastrnˇa´k A, Jansa P. Local eosinophilia in stroma of tumors related to prognosis. Neoplasma 1984; 31; 323–326. 6. Ohashi Y, Ishibashi S, Suzuki T et al. Significance of tumor associated tissue eosinophilia and other inflammatory cell infiltrate in early esophageal squamous cell carcinoma. Anticancer Res. 2000; 20; 3025–3030. 7. Deron P, Goossens A, Halama AR. Tumour-associated tissue eosinophilia in head and neck squamous-cell carcinoma. ORL J. Otorhinolaryngol. Relat. Spec. 1996; 58; 167–170.
2002 Blackwell Science Ltd, Histopathology, 41, 152–157.
8. Dessy E, Proto E, Fadda G, Pilato M. Reazione stromale, eosinofilia tessutale e mastocitosi nel carcinoma della laringe. Acta Otorhinolaringol. Ital. 1986; 6; 357–366. 9. Sassler AM, McClatchey KD, Wolf GT, Fisher SG. Eosinophilic infiltration in advanced laryngeal squamous cell carcinoma. Laryngoscope 1996; 105; 413–416. 10. Thompson AC, Bradley PJ, Griffin NR. Tumor-associated tissue eosinophilia and long-term prognosis for carcinoma of the larynx. Am. J. Surg. 1994; 168; 469–471. 11. Leighton SEJ, Teo JGC, Leung SF, Cheung AYK, Lee JCK, Hasselt CA. Prevalence and prognostic significance of tumor-associated tissue eosinophilia in nasopharyngeal carcinoma. Cancer 1996; 77; 436–440. 12. Looi L. Tumor-associated tissue eosinophilia in nasopharyngeal carcinoma. A pathologic study of 422 primary and 138 metastatic tumors. Cancer 1987; 59; 466–470. 13. Sta˚hle-Ba¨ckdahl M, Parks WC. 92-kd gelatinase is actively expressed by eosinophils and stored by neutrophils in squamous cell carcinoma. Am. J. Pathol. 1993; 142; 995–1000. 14. Ali S, Kaur J, Patel KD. Intercellular cell adhesion molecule-1, vascular cell adhesion molecule-1, and regulated on activation normal T cell expressed and secreted are expressed by human breast carcinoma cells and support eosinophil adhesion and activation. Am. J. Pathol. 2000; 157; 313–321. 15. Kodama T, Takada K, Kameya T, Shimosato Y, Tsuchiya R, Okabe T. Large cell carcinoma of the lung associated with marked eosinophilia. A case report. Cancer 1984; 54; 2313–2317. 16. Ferna´ndez-Acen˜ero MJ, Galindo-Galego M, Sanz J, Aljama A. Prognostic influence of tumor-associated eosinophilic infiltrate in colorectal carcinoma. Cancer 2000; 88; 1544–1548. 17. Iwasaki K, Torisu M, Fujimura T. Malignant tumor and eosinophils. 1. Prognostic significance in gastric cancer. Cancer 1986; 58; 1321–1327. 18. Nielsen HJ, Hansen U, Christensen IJ, Reimert CM, Bru¨nner N, Moesgaard F. Independent prognostic value of eosinophil and
2002 Blackwell Science Ltd, Histopathology, 41, 152–157.
23. 24. 25. 26. 27. 28.
mast cell infiltration in colorectal cancer tissue. J. Pathol. 1999; 189; 487–495. Pretlow TP, Keith EF, Cryar AK et al. Eosinophil infiltration of human colonic carcinomas as a prognostic indicator. Cancer Res. 1983; 43; 2997–3000. Ono Y, Fujii M, Kameyama K, Otani Y, Sakurai Y, Kanzaki J. Expression of matrix metalloproteinase-1 mRNA related to tissue eosinophilia and interleukin-5 gene expression in head and neck tumour tissue. Virchows Arch. 1997; 431; 305–310. Van Driel WJ, Hogendoorn PCW, Jansen F, Zwinderman AH, Trimbos JB, Fleuren GJ. Tumor-associated eosinophilic infiltrate of cervical cancer is indicative for a less effective immune response. Hum. Pathol. 1996; 27; 904–911. Van Driel WJ, Kievit-Tyson P, Van Den Broek LC, Zwinderman AH, Trimbos BJ, Fleuren GJ. Presence of an eosinophilic infiltrate in cervical squamous carcinoma results from a type 2 immune response. Gynecol. Oncol. 1999; 74; 188–195. Wardlaw AJ, Moqbel R, Kay B. Eosinophils: biology and role in disease. Adv. Immunol. 1995; 60; 151–266. Weller PF. Eosinophils: structure and functions. Curr. Opin. Immunol. 1994; 6; 85–90. Gleich GJ, Adolphson CR, Leiferman KM. The biology of the eosinophil leukocyte. Annu. Rev. Med. 1993; 44; 85–101. Resnick MB, Weller PF. Mechanisms of eosinophil recruitment. Am. J. Respir. Cell Mol. Biol. 1993; 8; 349–355. Silberstein DS. Eosinophil function in health and disease. Crit. Rev. Oncol. Hematol. 1995; 19; 47–77. Kowalski LP, Bagietto R, Lara JRL, Santos RL, Tagawa EK, Santos IRB. Factors influencing contralateral lymph node metastasis from oral carcinoma. Head Neck 1999; 21; 104–110. Ghiabi M, Gallagher GT, Wong DTW. Eosinophils, tissue eosinophilia, and eosinophil-derived transforming growth factor a in hamster oral carcinogenesis. Cancer Res. 1992; 52; 389–393. Tepper RI. The eosinophil-mediated antitumor activity of interleukin-4. J. Allergy Clin. Immunol. 1994; 94; 1225–1231.