Signs and symptoms of malignant parotid tumours: an objective assessment
D.S.Y. WONG
University Department of Surgery, Queen Mary Hospital, Pokfulam, Hong Kong

Introduction Patients and methods

Results Discussion References

Background: The clinical features that may be associated with malignancy in parotid tumours are well known. Classical teaching dictates caution in their presence but this raises a false alarm in many cases. Formal studies looking at these features are few. The aim of this article was to study quantitatively those features that provide a better prediction of malignancy. Methods: Clinical records of 186 consecutive patients treated for parotid tumours over a 12-year period were reviewed. Presence of suspicious clinical features and the final histology in each patient were noted. Results: The overall pick-up rate for malignancy, based on clinical features alone, was around 30%. Palpable cervical lymph nodes, facial nerve palsy, deep fixation and rapid enlargement of the tumour were significant parameters indicative of malignancy (p= 0.000 for all 4 parameters, chi-squared test). The risk of malignancy increased when multiple parameters were present together at the same time. Conclusion: Clinical features remained the most important single modality identifying malignancy in patients with parotid tumours. The logistic regression model allowed for simple clinical prediction of malignancy with improved sensitivity and much better specificity.

Keywords: clinical features, malignancy, parotid, signs, symptoms, tumour

J.R.Coll.Surg.Edinb., 46, April 2001, 91-95 

INTRODUCTION

The clinical features that may be present in a malignant tumour of the parotid gland are well known.^1-11 This knowledge is acquired through generations of experience but there have been scanty reports in the literature directed specifically to the study of these features.^12,13 Apart from their possible correlation with malignancy, the full implications of these features, therefore, have not been evaluated in detail. In particular, the following issues need to be addressed:

• Usefulness of the various clinical features
• Significance when several features are present simultaneously
• The relative importance of each feature
• A possible model for estimation of the risk in the clinical setting

The aim of this study, therefore, was to examine the signs and symptoms of malignant parotid tumours at initial presentation with a view to quantifying their importance so as to allow for a more effective clinical evaluation of the risk of malignancy.

PATIENTS AND METHODS

A list of the clinical symptoms and signs usually regarded as indicative of malignant tumours in the parotid was made up from a search of the literature.^1-13

The clinical records of patients treated for the diagnosis of a parotid tumour over the 12-year period, from January, 1986 to May, 1998, in the Division of Head & Neck /Plastic & Reconstructive Surgery of the Department of Surgery, University of Hong Kong Medical Centre, Queen Mary Hospital, were reviewed. Clinical features recorded in the records for each patient were then stored into a database, which also contained the final pathological diagnoses. With the statistical software package SPSS^® for Windows ^TM 7.5 (SPSS Inc., Chicago, Illinois), basic cross-tabulation inferences and correlations were performed, first with individual clinical features, and then for the case when several features were simultaneously present. The data was then fitted into the logistic regression model for identification of the independent significant features and for a model for calculation of the risk of malignancy based on the clinical features. Comparison of the efficacy of prediction made with our conventional approach based on the mere presence of suspicious features, and with the regression model, was then made.

RESULTS

One hundred and eighty-six patients were analyzed. There were 132 (71%) benign lesions, of which 15 (8.1%) were inflammatory and 117 (62.9%) neoplastic. Thirty-nine (21%) were malignant tumours.

The clinical features and their incidences are listed in Table 1, where results of evaluation of the individual clinical features are also shown. All patients with palpable cervical nodes had malignancy. No other clinical feature was likewise category specific and was present exclusively in either group. Facial nerve palsy was second in its importance in relation to a malignant outcome. Deep fixation and rapid recent enlargement of the tumour also were significant parameters, and were the other factors listed based on univariate analysis (Table 2).

Table 1: Individual clinical features and their significance

 Table 2: Individual clinical features: univariate analysis

Table 3 shows the numbers of clinical features present in the patients. All together 70 patients had some features suggestive of a malignancy in the parotid. The cumulative statistics from this table is shown in Table 4. A progressive trend in the probability of malignancy was obvious when more features were present simultaneously, irrespective of their particular nature. The probability of malignancy exceeded 0.5 when 2 or more features were present and this observation was also significant when tested by the chi-squared test (X²= 25.76, p=0.000).

Table 3: Occurrence of features indicative of malignancy in the benign and malignant groups

Table 4: Risks of malignancy in relation to the number of features present

When the various features were fitted into the multivariate logistic regression model, the symptom of rapid recent enlargement and the signs of cervical node enlargement, facial nerve palsy and deep fixation turned out to be significant factors, both with the forward as well as the backward stepwise selection methods. The constant (-2.22) and coefficients obtained with the forced entry mode are listed in Table 5. Taking a probability of >0.5 as at risk of an event occurring, and since all the features involved were dichotomous and indicator-variable coded as 0 (for absence) or 1 (for presence), a positive resulting ‘index’ (i.e. ‘Z’ in Table 5) from the calculation is indicative of a risk of malignancy¹⁴

Table 5: Calculation of risk of malignancy

The coefficients of only those features that are present are added to obtain S and then Z, where Z= [S + (-2.22)]. Z >0 implies significant risk (probability > 0.5) of malignancy.

DISCUSSION

Eighty percent of parotid gland tumours are benign of which eighty percent are pleomorphic adenomas.¹⁵ These present with an asymptomatic mass in the parotid region as do malignant tumours of the parotid in the early phase of growth.^4,8 Pre-operative biopsy in both situations is contraindicated because of the possibility of tumour spread and conversion of a potentially curable disease into one that is not.² Consequently, clinical examination is often the only assessment before operative treatment. This is particularly so in centres where fine needle aspiration cytology (FNAC) is not practised as a routine investigation before surgery.

A mass in the parotid region has to be a parotid tumour until proven otherwise. As malignant parotid tumours often are not diagnosed until they have been removed and assessed histologically, any parotid region mass may eventually turn out to be a malignant tumour.^16,17 A prior pre-operative knowledge of the malignant nature of a parotid tumour may be helpful in management in a number of ways. First, it allows proper patient briefing and reduces disappointments from unsuspected pathological diagnosis. Second, additional investigations such as FNAC and imaging studies can be arranged before surgery to increase the likelihood of surgical success. Third, the surgeon will be more likely to ensure excision with good margins, particularly when it becomes necessary to dissect into the deep lobe, when the deep aspect of the tumor is approached. Intra-operative frozen section, although not a routine procedure because of its known shortcomings, may provide further information and, thus, alter operative treatment.^18-21 The sentinel lymph nodes are also examined, and biopsied if indicated, when a tumour is suspected to be malignant.^1,2,9,22 This may, in turn, dictate the necessity for neck dissection. Although superficial parotidectomy will adequately remove most tumours, including malignant ones, full execution of the concept of agressive surgical treatment it has to be based on awareness of the malignant nature of a particular tumour.²² Furthermore, a helpful frozen-section diagnosis might indicate the tumor type and degree of malignancy and could avoid inadequate or over-extensive surgery.¹⁸ On the other hand, intelligent interpretation of clinical features may also reduce unnecessary anxiety to the patient.

In this study, the 15 inflammatory cases had been intentionally retained in the analysis. This is to simulate the real-life clinical situation where some inflammatory lesions are indistinguishable from tumours and, indeed, those were treated as such. The incidences of the various features are in good agreement with figures previously reported in the literature.^{1,3,6,9,11,23-25} In the present analysis, all patients with palpable cervical nodes had malignancy of the parotid. The incidence of metastatic nodes, overall, has been reported to vary from 6% to 33%.^{1,3,6,9,11,23-25} This wide range may be due to different referral patterns and inclusion criteria but cervical node involvement is an ominous sign. Pain is a more subjective symptom which may explain why it turned out not to be a significant factor. Pain is a component in inflammatory disorders but even in benign tumours, it may be present as a result of associated infection, haemorrhage or cystic enlargement.^2,5 Pain associated with a malignant tumour signifies nerve invasion and is a significant feature of malignancy. Though usually taken as pathognomonic of malignancy when present together with a parotid mass, rare occurrence of facial nerve palsy has been described in benign conditions.^7,26-35 A ‘hard’ consistency is similarly subject to observer variation as is skin involvement, unless the latter is overt. Incidentally, many of the strong indicators for malignancy are also known predictors of poor prognosis.^2-4,7,36,37

The present results show that clinical features are important in delineating malignant parotid tumours. Various investigatory tools, such as FNAC, sialography,⁴⁰ computerised tomography, magnetic resonance imagery, positon emission tomography and flow cytometric DNA analysis, have previously been studied and compared for their efficacy in this aspect and the consensus findings have suggested that clinical features as the most useful.^38-42 In this series, the overall efficacy of clinical features in identifying malignancies have been in the region of 30%.

However, it is evident that not only is it useful to recognize what feature is important, as emphasized in studies so far published, but also possible to weigh these factors, especially when more than one is present, to arrive at a more meaningful interpretation of the likelihood of malignancy in a particular clinical situation. At the same time, the isolated presence of a single or combination of clinical features would be difficult to interpret as most of these can be present in both benign and malignant conditions. A prediction model, therefore, is needed to address the problem. This is illustrated in Table 6, which compares the prediction based on the presence of the typical clinical features alone with that using the logistic regression model. Conventionally, we would be suspicious of malignancy whenever a feature is present and this gives rise to a large number of unnecessary false alarms (i.e. 70-10 = 60, in this series). Employing the regression model only slightly raised the sensitivity of the prediction but it greatly enhanced its specificity. This in turn means less unwarranted additional action to be taken both pre- and intra-operatively. Although the overall pickup rate remains at a less than ideal level, the value of careful evaluation of clinical features cannot be over-emphasized, particularly that it incurs no additional cost at all in this era of budget containment.

Table 6: Prediction with traditional model versus regression model

A systematic approach to the clinical assessment of a parotid tumour is recommended. In addition to the usual history taking and physical examination, a checklist of the symptoms and signs, as studied, is used. When any of the features are present, the corresponding coefficient is calculated (Table 5) and the sum total obtained. A positive value for the resulting index after the adjustment constant corresponds to a probability of higher than 0.5 of malignancy. When this increased risk of malignancy is present, FNAC with or without an imaging procedure becomes mandatory before surgery. In doubtful cases, intra-operative frozen section may also be employed, especially if the confirmation results in a change in the operative management. All too often the malignant nature of a parotid tumour becomes known only when the final pathology is available. However, in this situation, the tumour must have been so ‘early’ that, had a proper parotidectomy been performed, a prior knowledge of malignancy might not have resulted in a difference in the surgical management.

Despite the existence of a large number of publications on the parotid, articles directed specifically at studying the clinical features associated with malignant tumours are rare.^12,13 The results of this study emphasise the usefulness of clinical evaluation in diagnosing malignant parotid tumours. They also remind us that, despite technological advances in the investigatory work-up of patients available there is no substitute for careful clinical assessment of patients.

REFERENCES

1. Dunn EJ, Kent T, Hines J, Cohn I Jr. Parotid neoplasms: a report of 250 cases and review of the literature. Ann Surg 1976; 184: 500-6
2. Eisele DW, Johns ME. Salivary gland neoplasms. In: Bailey BJ (ed.). Head & Neck Surgery- Otolaryngology, 2nd edn. 1998, Philadelphia: Lipponcott-Raven Publishers
3. Sullivan MJ, Breslin K, McClatchey KD, Ho L, Farrior EH, Krause CJ. Malignant parotid gland tumors: a retrospective study. Otolaryngol Head Neck Surg 1987; 97:529-33
4. Pedersen D, Overgaard J, Sogaard H, Elbrond O, Overgaard M. Malignant parotid tumors in 110 consecutive patients: treatment results and prognosis. Laryngoscope 1992; 102:1064-9
5. Snyderman NL, Johnson JT. Salivary gland tumors. Diagnostic characteristics of the common types. Postgrad Med 1987; 82:105-12
6. Smith JA, Fleming WB. An approach to malignant parotid tumours. Aust N Z J Surg 1989; 59:317-20
7. Byrne MN, Spector JG. Parotid masses: evaluation, analysis, and current management. Laryngoscope 1988; 98:99-105
8. Kamal SA, Othman EO. Diagnosis and treatment of parotid tumours. J Laryngol Otol 1997; 111:316-21
9. Marandas P, Dharkar D, Davis A et al. Malignant tumours of the parotid: a study of 76 patients. Clin Otolaryngol 1990; 15:103-9
10. Banerjee AK, Ubhi CS, Pegg CA. Diagnostic approaches to parotid swellings. Br J Hosp Med 1994; 51:516-21
11. Rafla S. Malignant parotid tumors: natural history and treatment. Cancer 1977; 40:136-44
12. Phillips DE, Jones AS. Reliability of clinical examination in the diagnosis of parotid tumours. J R Coll Surg Edinb 1994; 39:100-2
13. Lam KH, Wei WI, Lau WF. Tumours of the parotid - the value of clinical assessment. Aust N Z J Surg 1986; 56:325-9
14. Norusis MJ. SPSS manual: Advanced statistics. 1993 , Chicago: SPSS Inc.
15. Shaheen OH. Benign salivary tumours. In: Kerr AE (ed.). Scott Brown’s Otolaryngology, 5th edn. Vol 5. 1987, London: Butterworth Publications
16. Fleming WB. ‘Benign’ pleomorphic adenoma of the parotid. Aust NJZ Surg 1987; 57:169-71
17. Eddey HH. Parotid tumours: a review of 138 cases. Aust NJZ Surg 1970; 40:1-14
18. Miller RH, Calcaterra TC, Paglia DE. Accuracy of frozen section diagnosis of parotid lesions. Ann-Otol Rhinol Laryngol 1979; 88:573-6
19. Remsen KA, Lucente FE, Biller HF. Reliability of frozen section diagnosis in head and neck neoplasms. Laryngoscope 1984; 94:519-24
20. Hillel AD, Fee WE Jr. Evaluation of frozen section in parotid gland surgery. Arch Otolaryngol 1983; 109:230-2
21. Dindzans LJ, Ven Nostrand AWP. The accuracy of frozen section diagnosis of parotid lesions. J Oto 1984; 13:382-6
22. Byers RM, Piorkowski R, Luna MA. Malignant parotid tumors in patients under 20 years-of-age. Arch Otolaryngol 1984; 110: 232-5
23. Kagan AR, Nussbaum H, Handler S et al. Recurrences from malignant parotid salivary gland tumors. Cancer 1976; 37:2600-4
24. Spiro RH, Huvos AG, Strong EW. Cancer of the parotid gland. A clinicopathologic study of 288 primary cases . Amer J Surg 1975; 30:452-9
25. Spiro RH. Salivary neoplasms: Overview of a 35-year experience with 2,807 patients. Head Neck Surg 1986: 8:177-84
26. La Venuta F, Flynn WF, Moore JA. Facial nerve paralysis secondary to benign parotid tumour. Arch Otolaryngol 1969; 90:603-4
27. Wilkie TF, White RA. Benign parotid tumour with facial nerve paralysis. Plast Reconstr Surg 1969; 43:528-30
28. Cimorra GA, Ferreira V, Martinex-Tello FJ. Spontaneous facial nerve paralysis associated with an ipsilateral benign parotid tumour. Plast Recontr Surg 1972; 50:523-31
29. Papangelou L, Alkalai K, Kyrillopoulou M. Facial nerve paralysis in the presence of a benign parotid tumour. Arch Otolaryngol 1982; 108:458-9
30. Lesser RW, Spector JG. Facial nerve palsy associated with Warthin’s tumor. Arch Otolaryngol 1985; 111:548-9
31. Koide C, Imai A, Nagaba A, Takahashi T. Pathological findings of the facial nerve in a case of facial nerve palsy associated with benign parotid tumor. Arch Otolaryngol Head Neck Surg 1994; 120:410-12
32. Delozier HL, Spinella MJ. Facial nerve paralysis with benign parotid mass. Ann Otol Rhinol Laryngol 1989; 98:644-7
33. Lesser RW, Spector JG. Facial nerve palsy associated with Warthin’s tumour. Arch Otolaryngol 1985; 11:548-9
34. Mirza N, Crumley R. Facial paralysis in a benign osseous parotid tumor: a case report. Otolaryngol Head Neck Surg 1993; 108: 367-71
35. Wilkie TF, White RA. Benign parotid tumor with facial nerve paralysis. Case report. Plast Reconstr Surg 1969; 43:528-30
36. Spiro IJ, Wang CC, Montgomery WW. Carcinoma of the parotid gland. Analysis of treatment results and patterns of failure after combined surgery and radiation therapy. Cancer 1993; 71:2699-705
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; 82:1217-24
38. Deans GT, Briggs K, Spence RA. An audit of surgery of the parotid gland. Ann R Coll Surg Engl 1995; 77:188-92
39. Takashima S, Takayama F, Wang Q, Kurozumi M, Sekiyama Y, Sone S. Parotid gland lesions: diagnosis of malignancy with MRI and flow cytometric DNA analysis and cytology in fine-needle aspiration biopsy. Head Neck 1999; 21:43-51
40. Morgan RF, Saunders JR Jr, Hirata RM, Jaques DA. A comparative analysis of the clinical, sialographic, and pathologic findings in parotid disease. Am Surg 1985; 51:664-7
41. McGuirt WF, Keyes JW Jr, Greven KM, Williams DW 3rd, Watson NE Jr, Cappellari. Preoperative identification of benign versus malignant parotid masses: a comparative study including positron emission tomography. Laryngoscope 1995; 105:579-84
42. Freling NJ, Molenaar WM, Vermey A, Mooyaart EL, Panders AK, Annyas AA, Thijn CJ. Malignant parotid tumors: clinical use of MR imaging and histologic correlation. Radiology 1992; 185:691-96

Copyright date: 11 January 2001

Correspondance: Dr David S. Y. Wong, Division of Head & Neck/ Plastic & Reconstructive Surgery, Department of Surgery, University of Hong Kong Medical Centre, Queen Mary Hospital, Pokfulam, Hong Kong

E-mail: sywong@ha.org.hk

©2001 The Royal College of Surgeons of Edinburgh, J.R.Coll.Surg.Edinb.