To the Editor.
—We found the article by Hamrick and Murgo1 interesting and informative. Prediction of carcinomatous involvement of the bone marrow without performance of a bone marrow biopsy remains a problem for clinicians dealing with cancer. If less invasive means of determining bone marrow involvement could be found, it would save our patients from having to undergo the discomfort of a bone marrow biopsy.In support of Hamrick and Murgo's findings, we would like to point out a similar investigation2 looking for predictors of bone marrow metastases from malignant diseases, including small-cell lung carcinoma. Stepwise discriminate analysis in that report2 found four parameters that strongly correlated with bone marrow involvement: the presence of a leukoerythroblastic blood pattern on peripheral blood smear, a serum lactate dehydrogenase concentration over 500 U/L, a platelet count under 100 × 109/L (100 000/mm3) and bone pain. Four parameters