INTERMETHOD VARIATION IN BONE HISTOMORPHOMETRY - COMPARISON BETWEEN MANUAL AND COMPUTERIZED METHODS APPLIED TO ILIAC BONE BIOPSIES

The intermethod variation in the measurement of basic bone histomorphometric parameters was evaluated on 100 undecalcified transiliac bone biopsies. Two contiguous samples were taken from 50 patients (33 females; 17 males; mean age: 52 .+-. 19 years) for diagnostic purposes. The diagnoses were osteoporosis (n = 38), renal osteodystrophy (n = 18), primary hyperparathyroidism (n = 16), osteomalicia (n = 12), metastatic bone disease (n = 2), thyrotoxic bone (n = 2), fluorosis (n = 2), and 10 biopsies were considered as "normal" bone. Trabecular bone volume (TBV) was measured with both a manual integrating eyepiece and an automatic (QUANTIMET 720-Cambridge Instruments, Cambridge, England) method. Trabecular resorption surfaces (TRS), trabecular osteoid surfaces (TOS), and volume (TOV) were measured with both a manual and a semiautomatic (VIDEOPLAN-Kontron, Munich, West Germany) method. The calcification rate (CR) was measured with both a manual and a semiautomatic method in eight cases after double labeling with tetracycline. Inter- and intraobserver variations were always lower with the automatic and semiautomatic methods than with the manual method, except for TOV. For all the parameters there was a highly significant correlation between manual and computerized methods (0.98 > r > 0.90). For TBV and CR no significant difference was noted, but for TBV the QUANTIMET appeared more sensitive, that is, better able to detect low values of the structure to be measured. For TRS, the manual method underestimated low values and appeared less sensitive than the semiautomatic method. For the 100 biopsies, the VIDEOPLAN underestimated the osteoid parameters by 13% for TOS and 26% for TOV. These underestimations could be corrected by modifying measuring procedures or by using higher magnification. In conclusion, TRS and osteoid parameters measured with one method cannot be compared with the ones obtained with another, without considering the intermethod variations that depend on the diagnosis. The automatic method is preferred for TBV; the semiautomatic method for TRS, TOS, and CR; and the manual method for TOV.