Abstract: Probability Predicting Tool for Identifying Incidence and Severity of Pancytopenia as a Result of Megaloblastic Anemia

Anuradha Vutukuru and Attili Vs Suresh

Background: One of the most common etiology in the diagnosis of pancytopenia in physician practice is megaloblastic anemia. However there is significant confusion associated with marrow megaloblastic features, which need not always because of pancytopenia. They may be just co-incidental findings. Its often difficult to establish that the peripheral pancytopenia is related to marrow megaloblastic features, as Vitamin B12 and folate are often normal in these cases [because of prior treatment]

Aim: This study was conducted to develop a probability predicting system for possible incidence and severity of pancytopenia as a result of megaloblastic anemia.

Materials and Methods: This is a retrospective analysis conducted at a tertiary care center with approximately 2,000 new cases of megaloblastic anemia. We Hypothesized age, duration of symptoms, Mean Corpuscular Volume [MCV], nutritional status, B12 and folate levels, underlying illness and response to therapy as possible factors associated with pancytopenia of megaloblastic anemia. Receiver operating characteristic (ROC) curves were drawn to predict the cutoff values for risk factors, and a final scoring system was developed with sensitivity and specificity data.

Results: A total of 458 patients with pancytopenia and marrow findings of megaloblastic anemia were analyzed. Based on ROC analysis, following cutoff values were selected: age > 40 years or <20, Folate <30% lower limit, B12 any value close to lower limit of reference laboratory value, Mean Corpuscular Volume>110 fl, duration of symptoms more than 18 weeks, Serum Albumin <2.5 gm/dl [taken as marker for nutrition]. The remaining factors were indicated as present or absent. A score of 1 was assigned for the above factors if they were present. For patients, the final score of 2 or less there is 22% probability of having pancytopenia while patients having score of 6 or more have 89% probability. Similarly when the Mean Corpuscular Volume recovered/reduced by 8 fl at week 3, patients have positive predictable recovery pattern. Those who does not have such recovery by week 3, the etiology of pancytopenia is unlikely to be megaloblastic anemia. And they need further evaluation