Combined radiology and cytology in the diagnosis of bone lesions : a study of 494 patients

Abstract

The aim of the present study was to assess the validity of combining radiology and fine needle aspiration cytology in the diagnosis of bone lesions.

Patients and Methods Radiological, cytological and histopathologic findings in 494 cases diagnosed 1993 through 1998 were collected from the original reports. The study included four overlapping series: 1) 52 consecutive high grade osteosarcomas, 2) 83 consecutive myelomas/lymphomas, 3) 29 unselected lesions analysed for sample representativeness and 4) another 370 consecutive unselected lesions analysed for diagnostic accuracy. To compare radiology, cytology and also histopathology, the lesions, were categorised into four groups: 1) sarcoma, 2) metastasis, myeloma or lymphoma, 3) benign tumour and 4) non-neoplastic/ normal lesion.

Osteosarcoma The 52 osteosarcoma cases were radiologically sub-grouped into typical and atypical based on location and appearance. 32 of 33 typical cases were cytologically assessed as osteosarcoma and 1 as sarcoma NOS. The remaining 19 atypical cases were cytologically assessed as osteosarcoma (6), as other sarcomas (9), as benign (1) and as nondiagnostic (3). In radiologically typical osteosarcoma cytology provides safe diagnostic confirmation. However, in radio-logically atypical osteosarcoma open or core biopsy should be considered before decisions about therapy.

Myeloma/lymphoma In a series of 40 myelomas and 43 lymphomas, all lesions were radiologically classified as aggressive without specific features. All 40 myelomas and 41/43 lymphomas could be diagnosed by cytology and in 70 (32 myelomas, 38 lymphomas) the diagnosis was corroborated by immunocytochemistry. In conclusion, skeletal myeloma and lymphoma are easily detected by radiology and reliably diagnosed by cytology and immunocytochemistry.

Representativeness In a series of 29 consecutive, unselected patients with a bone lesion, 3 aspirates from each lesions were taken to assess sample representativeness. The 87 aspirates were reviewed randomly by two cytologists unaware of clinical data Among 29 lesions, 13 and 12, respectively, were assessed by the two cytologists as homogenous, i.e. the 3 aspirates of each lesion exhibited the same cellular picture. In the remaining lesions, heterogeneity almost exclusively pertained to the mixture of conclusive and non-conclusive aspirates.

Two alternative diagnoses were suggested in 1 case by each cytologist. Comparison of the two cytologists assessments disclosed concordance in 21 cases (63 aspirates). In the remaining 8 cases (24 aspirates), interobserver difference was mainly due to different ratios of conclusive v. non-conclusive aspirates. Notably, only the analysis of one and the same aspirate resulted in two different diagnoses. A correct diagnosis was given by the cytologists in 22 and 23 cases, respectively, incorrect in 2 and non-conclusive in 5 and 4. The study suggests that true tumour heterogeneity is rare. The interobserver difference and diagnostic difficulties should mainly be attributed to the "blind", random approach of the analysis.

Combined radiology and cytology Analysis of the diagnostic accuracy of combined radiology and cytology was based on 399 unselected patients with a bone lesion. In 385 cases a cytological diagnosis could be established, whereas 14 cases. were nondiagnostic. Comparison of radiology and cytology showed that 279 (72 %) were diagnostically compliant according to category, the remaining 106 non-compliant (28 %). Histopathological analysis was done in 134 cases (86 compliant, 48 noncompliant). The remaining 251 were managed by radiology and cytology only. At follow up 2/251 proved to have metastasis despite normal cytology. Altogether, 19 cases were misdiagnosed, 2 among the 279 compliant cases (0.7 %) and 17 among the 106 non-compliant (16 %). However, the latter because of non-compliance were managed also by either open biopsy, surgery or follow up, disclosing that not one single patient had been put at risk.

The results of the present study suggest that in an unselected series of patients with a bone lesion 72 % can be reliably diagnosed by combining information from radiography and cytology. Among the remaining 28 %, the patients requiring further diagnostic work up are easily identified. The strategy described entails a low risk of false diagnosis, which is comparable to that obtained by consistently resorting to histopathology for microcharacterisation.