Time between initial presentation and development of distant lesions.
A, Disease-specific survival for patients with metastasizing pleomorphic adenoma; B, disease-free survival for patients with metastasizing pleomorphic adenoma (Kaplan-Meier analysis).
A, Disease-specific survival as a function of the presence of lesions in a single site or multiple distant sites; B, disease-specific survival as a function of the length of time between initial tumor presentation and detection of metastasis (Kaplan-Meier analysis).
A 2-[18F]fluoro-2-deoxy-d-glucose positron emission tomography scan of a 25-year-old patient investigated for recurrence of a parotid pleomorphic salivary adenoma at the skull base (A) (arrows), revealing an unsuspected pulmonary lesion (B) (arrows), which biopsy results proved to be histologically identical to the preceding pleomorphic salivary adenoma.
Nouraei SAR, Ferguson MS, Clarke PM, Sandison A, Sandhu GS, Michaels L, Rhys-Evans P. Metastasizing Pleomorphic Salivary Adenoma. Arch Otolaryngol Head Neck Surg. 2006;132(7):788-793. doi:10.1001/archotol.132.7.788
Copyright 2006 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2006
To address questions about the etiology, behavior, optimal treatment, and prognosis of metastasizing pleomorphic adenoma (MPA), we undertook a review of the literature (1953-2005) and constructed a virtual series of all identified cases of MPA, metastatic lesions that are very occasionally identified in patients with a history of pleomorphic salivary adenoma and, on detailed pathological evaluation, found to exhibit all the histological hallmarks of the preceding benign lesions.
A review of the English-language literature between 1953 and 2005 using MEDLINE, secondary references identified from bibliographies of pertinent articles, and a further case from one of our institutions.
A virtual case series was constructed and quantitatively analyzed. Forty-two patients with an average age of 33 years were identified. There were 20 male and 22 female patients. There was an overwhelming history of incomplete surgery for pleomorphic salivary adenoma. Most patients had locoregional recurrences before metastasis, and the mean presentation-to-metastasis latency was 16 years. Bone was the most common site for metastases (45%), followed by the head and neck (43%) and lung (36%). There was significant morbidity and mortality from distant disease, with 5-year disease-specific and disease-free survival of 58% and 50%, respectively. Developing distant lesions within 10 years of the primary tumor and presence of metastases in multiple sites were independent predictors of survival on Cox regression analysis. Metastasectomy conferred significant survival advantage over nonoperative treatment (log-rank analysis, P<.02). Chemotherapy and radiotherapy were of limited value.
Meticulous surgery is crucial in preventing MPA. Metastatic disease carries significant morbidity and mortality and should be treated surgically when feasible.
Pleomorphic salivary adenomas (PSAs) are the most common neoplasm affecting the salivary glands.1- 3 They comprise both epithelial and mesenchymal structures and are considered biologically benign.4,5 Occasionally, the epithelial component, in isolation or in conjunction with the stroma, undergoes malignant transformation, giving rise to carcinoma ex pleomorphic adenoma or carcinosarcoma, respectively.3,6,7 These are both aggressive malignant tumors. Very occasionally however, metastatic lesions are identified in patients with a history of PSA, which, on detailed pathological evaluation, are found to exhibit all the histological hallmarks of the preceding benign lesions. This enigmatic entity has been termed the metastasizing pleomorphic adenoma (MPA).8 In this study, we undertook a review of the literature and constructed a virtual series of all identified cases of MPA reported in the English-language literature between 1953 and 2005, in an attempt to address questions about etiology, behavior, optimal treatment, and prognosis for this condition.
We conducted a literature search using MEDLINE, accessed via the National Library of Medicine PubMed interface (http://www.ncbi.nlm.nih.gov/pubmed), searching for articles relating to MPA written in English. We used the following search string: ([pleomorphic OR mixed] AND [benign OR adenoma OR parotid OR parotids OR submandibular OR sublingual OR salivary] AND [metastasising OR metastasizing OR metastasis OR metastasise OR metastasize OR metastatic OR metastases]). We also used the “Related Articles” feature of PubMed to identify further references of interest within the primary search. These references were obtained, and from their bibliographies, pertinent secondary references were also identified and acquired. The process was repeated until no further new articles could be identified.
Five reviewers (3 surgeons and 2 head and neck pathologists) independently assessed all of the articles. References prior to the classification of salivary gland tumors by Foote and Frazell2 in 1953,2 articles giving insufficient clinical or histological information, cases reported as benign MPA, which, on further review were found to have been reports of local recurrences, or cases in which the primary tumor and the metastasis were reported to have been histologically different were excluded from further analysis. Of the remaining articles, cases in which the reported histological features were in accord with the criteria established by the World Health Organization for the diagnosis of this condition8 were included. This provided us with 41 cases from the literature, to which we added a further previously unreported case from one of our institutions.
Details of initial presentation and treatment, histopathological condition, local recurrence, metastasis, and survival data were extracted from case reports fulfilling the inclusion criteria. Information was also gathered about the methods used for pathological evaluation of these lesions. This information was constructed into a virtual case series in the same way as a retrospective chart review study and analyzed using SPSS 12.0 for Windows (SPSS Inc, Chicago, Ill). Table 1 provides information about all the cases included in this series, and information on some of those excluded from further analysis is given in Table 2.
Data were presented either as means ± SD or percentages when appropriate. Log-rank analysis was performed to assess the impact of potential prognostic variables on survival. A Cox proportional hazards ratio analysis was performed to identify independent predictors of survival. Survival data were illustrated using the method of Kaplan and Meier, and P<.05 was considered statistically significant.
Forty-two cases fulfilling the inclusion criteria were identified between 1953 and 2005. There were 20 male and 22 female patients, and the mean ± SD at initial presentation was 32.9 ± 18.5 years (range, 8-73 years). The most common site of the initial tumor was the parotid gland (74%), followed by minor salivary (17%) and submandibular (10%) glands. Initial treatment consisted of local excision (64%), enucleation (12%), superficial parotidectomy (17%) and total parotidectomy (5%). One patient was treated with primary chemotherapy.28
Most patients in this series (81%) had at least 1 local recurrence prior to detection of a distant metastasis. The mean ± SD time from presentation to recurrence was 5.0 ± 4.9 years. In cases with 3 or more recurrences (19%), there was generally an escalation toward radical local surgery, including, for example, resection of the petrous temporal bone, the external auditory meatus, and the posterior mandible in a patient with 2 previous parotid enucleations and a total parotidectomy.30
The mean ± SD time from initial presentation to the detection of metastases for patients who had no intervening local recurrences was 12.3 ± 8.6 years. For patients with a history of local recurrence, this was 16.9 ± 13.3 years. The difference in the presentation to metastasis latency between the groups fell outside statistical significance (P>.40, Mann-Whitney test). Figure 1 illustrates the range of time from initial presentation to metastasis.
In 7 patients (17%), detection of a metastasis was the result of medical evaluation of a locally recurrent PSA. In 3 cases (7%), metastases were discovered while the patient was being investigated for an unrelated disease. A further 25 patients (60%) presented symptomatically (Table 3), and the mode of detection was not clearly documented in 7 cases (17%). Mode of detection did not influence patient survival (P>.10, log-rank analysis).
Bone was the most common site for metastasis (45%), followed by the head and neck (43%), lungs (36%), and abdominal viscera (10%). Within the head and neck, there were 7 metastases (17%) to regional lymph nodes, and 9 tumor deposits (21%) were detected in nonlymphatic extracranial tissues such as skin or paranasal sinuses. There were 2 cases of intracranial metastases. Further information about the location of metastases is provided in Table 4.
Surgery was the mainstay treatment of the metastases irrespective of the distant site. Surgical treatment most commonly consisted of local tumor resection, and 6 patients received adjuvant radiotherapy. Six patients were treated nonoperatively. Of these, 3 were treated with primary radiotherapy, 2 were observed, and 1 received chemotherapy. Metastasectomy conferred a significant survival advantage on univariate log-rank analysis (hazard ratio, 0.21; 95% confidence interval, 0.02-0.85; P<.01) over nonoperative treatment.
The mean ± SD follow-up time from the detection of metastases was 3.2 ± 3.3 years (range, 1-16 years). The disease, once established in a distant site, was associated with significant morbidity and mortality. The 5-year disease-specific and disease-free survivals were 58% and 50%, respectively (Figure 2). Examples of morbidity associated with this condition included spinal cordotomy for intractable pain secondary to lumbar spine disease11 and sacral resection with excision of the sacral plexus, also leading to paralysis and double incontinence.30 Independent prognostic factors identified with Cox regression analysis were whether metastases were present in a single site or multiple sites (Figure 3A), and the interval between initial presentation of a local PSA and detection of a distant metastasis. Patients presenting with metastatic lesions within 10 years of their initial primary tumor presentation had a significantly worse prognosis compared with those patients whose metastases were detected more than 10 years from their initial primary PSA presentation (Figure 3B).
In all cases, the original specimen and the metastasis were evaluated by light microscopy and staining. A proportion of these cases were additionally evaluated by immunohistochemistry, flow cytometry, and DNA analysis. Features of malignancy, including increased mitotic activity, areas of necrosis, infiltrative pattern, and cellular atypia were actively sought but not identified in any of the primary lesions or distant metastases included in this series. In 8 cases (19%), the initial pathological diagnosis was revised in light of further clinical information about the history of a PSA being made available to the pathologist in retrospect. In all but 1 of these cases, the definitive pathological diagnosis was reached through staining and light microscopy alone. Details of the pathological evaluation of these patients and the initial pathological diagnoses are provided in Table 5.
In this study, we retrospectively reviewed all cases of MPA that could be identified in the English-language literature between 1953 and 2005 and included a hitherto unreported case from our institution. While there are limitations to this study due primarily to the relatively small sample size resulting from the rarity of the condition, it was nevertheless possible to make important inferences about etiology, behavior, and outcome of patients presenting with this enigmatic condition.
Metastasizing pleomorphic adenoma occurred on average 16 years following the initial treatment of a benign pleomorphic adenoma. The primary tumor arose most commonly from the parotid gland, followed by the minor salivary and submandibular glands. This distribution corresponds with that of PSAs in general.1
Incomplete tumor clearance has been strongly linked with local recurrence.48 In this series we also noted an overwhelming association between incomplete surgical excision of the primary lesion and distant metastases. This further supports the current paradigm that meticulous tumor resection with adequate margins should be used and that enucleation should play no role in the management of a benign pleomorphic adenoma.
The most recent World Health Organization classification of salivary gland neoplasms categorizes an MPA as a malignant epithelial neoplasm and defines it as “a histologically benign pleomorphic adenoma that inexplicably manifests local or distant metastasis.”8 The fact that a tumor with a benign histological appearance can present with metastases and behave in a clinically malignant manner is a highly unusual but not a unique scenario. Giant cell tumors of the bone can also behave in exactly the same way.49 Our data confirm that even though based on current methods of histological diagnosis, MPA cannot be differentiated from a benign pleomorphic adenoma, its natural history is that of an aggressive malignant entity. Further research is needed to identify the biological reasons behind the observed aggressive behavior of this tumor.
The majority of metastases in this series presented symptomatically. On the other hand, most of the asymptomatic lesions were discovered as part of the diagnostic workup of a local recurrence. It is at present impossible to determine which of the locally recurrent PSA lesions have the potential to give rise to an MPA and therefore, investigation of a locally recurrent PSA should, in our view, include a search for distant metastases. Positron emission tomography is our investigation of choice (Figure 4). While this choice of imaging is based on limited experience with MPA itself, it is used on the premise that local PSA lesions are identified,50 and by extrapolation metastatic lesions could be identified, with positron emission tomography. We also advocate, in view of the length of time between the initial presentation and metastasis, that patients with intraoperative tumor spillage, incomplete tumor resection, or a local recurrence, should have long-term follow-up.
Presence of metastases in multiple locations was an independent negative prognostic factor, as was the occurrence of metastasis within 10 years of initial presentation. We wonder if the latter observation reflects differences in host susceptibility—metastatic PSA has been shown to progress rapidly with immunosuppression31—or a more aggressive tumor biological behavior.
Metastasectomy was the mainstay of treatment, and a minority of patients were treated nonoperatively. Surgical treatment conferred a significant survival advantage over nonoperative treatment on log-rank analysis (P<.01). Chemotherapy and primary radiotherapy were not very effective in this disease.
This condition, once established, is associated with significant disease-specific morbidity and mortality caused by distant-site disease progression and postmetastasectomy recurrence. Five-year disease-specific and disease-free survival rates were 58% and 50%, respectively. This further emphasizes the importance of meticulous surgical technique in treating the primary disease to prevent tumor seeding of the surgical field.
It is clear from this series that an important factor in reaching the histological diagnosis is communication of the clinical history. This lesion, even when occurring at a distant site, bears the characteristic hallmarks of a benign pleomorphic adenoma and should be identifiable on histological examination. However, in 19% of cases in this series the initial pathological diagnosis was revised, and the correct diagnosis was only reached after further communication between the referring clinician and the pathologist. It is therefore imperative that such lesions are discussed within a multidisciplinary team, and the diagnosis of an MPA is considered when a history of PSA exists.
In conclusion, MPA is a rare condition in which the etiology remains incompletely understood and the histological appearance, currently indistinguishable from a benign pleomorphic adenoma, is at great variance with its clinically malignant behavior. Complete meticulous excision of the primary lesion may reduce the likelihood of distant spread. Development of metastases within 10 years of initial presentation and the presence of disease in multiple distant sites significantly reduce survival. Once established, MPA causes significant morbidity and mortality, and metastasectomy, when technically feasible, appears to be the optimal management.
Correspondence: S. A. R. Nouraei, MBBChir, Department of Otolaryngology, Charing Cross Hospital, London W6 8RF, England (RN@cantab.net).
Submitted for Publication: June 8, 2005; final revision received October 24, 2005; accepted November 28, 2005.
Author Contributions: Drs Nouraei and Ferguson contributed equally to the manuscript.
Financial Disclosure: None reported.
Previous Presentation: This study was presented at the Summer Meeting of the British Association of Plastic Surgeons; July 7, 2005; Windsor, England.