Reliability of symptoms to determine use of bone scans to identify bone metastases in lung cancer: prospective study

Introduction

Based on the hypotheses that most skeletal metastases in lung cancer are clinically symptomatic, that the incidence of bony metastases in early stages is low, and that bone scintigraphy has a sensitivity of nearly 100%, leading professional societies recommend diagnostic skeletal imaging depending on clinical symptoms.1 2 No study has assessed the significance of skeletal symptoms as a criterion for skeletal imaging in patients with lung cancer since 1991.3 But in the intervening period gamma camera technology has been considerably refined and more sensitive methods such as magnetic resonance imaging have become available for skeletal imaging.

We redetermine the role of symptoms and serum concentrations in detecting bony metastases in lung cancer and reassess the accuracy of bone scans for screening.

Participants, methods, and results

From September 1999 to September 2001 we recruited 153 consecutive patients at University Hospital Ulm. We included patients based on cytological or histological evidence of lung cancer returned no more than 10 days before entry into the study. Of these, 121 (79%; 88 men and 33 women; median age 66, range 40-83 years) agreed to participate. Exclusion criteria were a history of malignant disease, pregnancy, and age less than 18 years. All patients gave written informed consent. Diagnosis was non-small cell lung cancer in 84 patients and small cell lung cancer in 37 patients. We questioned and examined all patients about skeletal complaints. Physical examination included percussion, compression, flexion, extension, and rotation of the vertebral column and extremities and evaluations of patients' neurological status. We also measured serum calcium and alkaline phosphatase concentrations. New skeletal symptoms within the previous six months were judged as suspicious for bony metastases.

We did bone scans blinded to the history and findings of the physical examination. The combined results of magnetic resonance imaging of the vertebral column and patients' subsequent clinical course were the ideal for identification of bony metastases.

We found skeletal metastases in 40 patients (33%). Incidence was nearly identical at 33% (28) in patients with non-small cell lung cancer and 32% (12) in those with small cell lung cancer. These patients had normal serum alkaline phosphatase and calcium concentrations. Three quarters (91) of patients had symptoms. In only 19% (23) of patients with symptoms did the location of metastases correspond to the symptoms. Routine bone scans correctly identified skeletal metastases in 29 patients (sensitivity 73%; 95% confidence interval 56% to 85%). Bone scans were correctly negative in 80 of 81 patients (specificity 99%; 93% to 100%). If bone scans were done in only the 91 patients reporting skeletal complaints, the sensitivity would have been reduced to 53%. A further restriction of the method to those 23 patients with suspicious complaints would have resulted in a further reduction in sensitivity to 20% (8 patients).

Comment

Only a small proportion of bone metastases can be recognised based on clinical symptoms or by increased serum alkaline phosphatase or calcium levels. Because of the high incidence of arthritic complaints,4 guidelines mandating bone scans only in patients with skeletal complaints often result in coincidental discovery of bony metastases (figure). Restricting bone scans to patients whose skeletal complaints were clinically suspicious for metastatic disease, however, would have resulted in an unacceptably low sensitivity of 20%. Considering the high incidence of bone metastases at initial diagnosis, this might lead to a dramatic increase in the number of patients undergoing futile surgery or neoadjuvant chemotherapy.

Bone scan showing a 49 year old white male in good general condition (Eastern Cooperative Oncology Group performance status 0) diagnosed with non-small cell lung cancer (cT3N2M0). He reported pain in the shoulder joints of two years' duration. Physical examination failed to show pain on percussion, pressure, or movement of the vertebral column. His complaints were correctly interpreted by the examiner as being arthritic in origin. A bone scan got despite the examiner's interpretation of the patient's complaints identified an asymptomatic metastasis localised to the fourth thoracic vertebra. The correct interpretation of the patient's symptoms and the resulting decision to defer skeletal scintigraphy would have represented false negative findings and would have resulted in the patient undergoing neoadjuvant treatment