New-onset proximal muscle pain or stiffness in subjects older than 50 years are most often caused by polymyalgia rheumatica (PMR), but polymyalgic pain can also be present in a wide variety of disorders, such as malignant neoplasms, endocarditis, different vasculitic disorders, and connective tissue diseases (for review see reference 1).
Based on clinical experience, physicians believe PMR to be associated with malignancy (2). However, no increased risk of malignancy has been found in systematic studies examining patients with PMR (3–6). A prospective study of 287 patients with PMR or temporal arteritis (TA) disclosed only 1 case of malignancy in a patient with TA, but none in those with PMR (5). The diagnostic criteria used for PMR are often based on the exclusion of disorders that can present with similar symptoms, including rheumatoid arthritis and malignancy (7, 8). It is possible that the negative finding of malignancy in the studies of PMR (3–5) is due to a prior selection of patients presenting with polymyalgic symptoms. In a small study, we tried to overcome this potential bias by including patients with new-onset polymyalgic symptoms suspected to have PMR before they underwent a more extensive examination.
The study, approved by the local ethics committee, was performed during a 24-month period (1995–1997) in the county of Vest-Agder in Norway with one rheumatology referral unit and a population of 150,426. In response to our mailed request to general practitioners in the community and doctors in other hospital departments, 31 women and 19 men (mean age 71.3 years) were referred with new-onset polymyalgic symptoms suspected of being associated with PMR. Within 2 weeks, the patients were assessed at our rheumatology department by experienced rheumatologists, and a number of baseline investigations were performed routinely. These included laboratory tests (e.g., hematology, liver function tests, creatinine, rheumatoid factor, antinuclear antibodies, serum electrophoresis), chest radiograph, abdominal ultrasound, and corticosteroid stress test (15 mg prednisolone on 3 consecutive days). In some patients a more extensive screening was performed, according to the judgment of the examining doctor, including abdominal computed tomography (CT), bone scan, bone marrow examination, and biopsy of the temporal artery.
As shown in Table 1, PMR was diagnosed in 40 patients, malignancy in 5, TA in 2, and miscellaneous disorders in 3. The Bird criteria (9) were fulfilled in all patients except 2 diagnosed as having PMR. Thirty-five of the 40 patients with PMR, both patients with TA, 1 patient with myelodysplasia, and the patient with lymphoma fulfilled the criteria by Healey et al (10). The rest of the patients did not fulfill these criteria.
|Diagnoses||No.||No. fulfilling Bird criteria|
|Unknown primary tumor with liver metastases||1||1|
Within the malignancy group, 1 patient had lymphoma, which was diagnosed by enlarged neck lymph nodes found by clinical examination prior to the corticosteroid stress test. Among the other patients, only 1 with myelodysplasia responded initially to the corticosteroid test, which led to a delay of several months in diagnosing this patient. The 3 other patients did not respond to the corticosteroid test. In the patient with prostate cancer, a greatly elevated alkaline phosphatase was found. Bone scan showed several lesions suggesting metastases, and on clinical examination of the prostate, malignancy was suspected. Another patient was found to have multiple liver metastases on abdominal CT scan, which were confirmed by biopsy despite a previous negative ultrasound examination. In the patients with myelodysplasia, further examination of the bone marrow was needed.
The observed frequency of malignancy in these patients, initially suspected to have PMR, was compared with the frequency of expected malignancy in the Norwegian population over the same 2-year period (1995–1997), adjusted for age and sex, given by the national cancer registry in Norway. The frequency of malignancy in the Norwegian population of 1,302,633 inhabitants over age 50 years was 1.3% in 1993. Giving the same age and sex distribution to the Norwegian population as found in our patients, the expected frequency in the normal population would have been 1.6%. A two-tailed Fisher's exact test (cross tabulation) revealed a highly statistically significant difference (P = 0.0013) between the expected occurrence of malignancy of 1.6% and the observed 10% in our cohort of polymyalgic patients.
The obvious main limitation of our results is the small number of patients investigated, and they should be interpreted with care. However, we do believe that our data raise justified concern that malignancy indeed may occur more frequently in patients suspected initially to have PMR than in the general population. The experience of many physicians may, therefore, be true.
The present study was not designed as a case control study, which might be criticized as a limitation. Instead, we compared the frequency of malignancy in our examined patients with the official malignancy incidence data given by the national cancer registry in Norway. In fact, this is statistically a more conservative model of detecting a difference between groups than is establishing a control group of randomly selected subjects from the general population.
Based on our own experience, we believe that a thorough initial medical examination in patients with suspected PMR would exclude most of the cases with an underlying pathology. However, we recommend screening for malignancy in those patients where no other obvious diagnosis can be found.
Our hypothesis, that the prevalence of malignancy is increased in patients with suspected PMR when compared to the general population, needs to be confirmed in prospective studies. Inclusion criteria should include unselected patients with new-onset polymyalgic symptoms.