Modified International Prognostic Score (IPS) for Predicting the Need for Bone Marrow Biopsy (BMB) in Hodgkin's Lymphoma (HL).

BMB is frequently performed during the staging process of patients (pts) with HL. Previous data suggested routine use of this procedure may be unnecessary. The Ann Arbor classification is currently used to detect pts requiring BMB, despite sensitivity (sen) of 100%, specificity (spec) is only 40%. We sought to determine whether a modified IPS [Hasenclever and Diehl (1998)] would yield greater spec while maintaining good sen.

We retrospectively reviewed charts of 1215 histologically proven HL pts from Jan 2000-Dec 2008 at Cleveland Clinic Taussig and Moll Cancer Centers. Of 1215 pts, 1089 had BMB (90%). 876 pts were included in our study. 213 were excluded due tor un-interpretable BMB or missing data. Each component of IPS was equally weighted such that one point is assigned to each positive variable [male sex, age ≥45, albumin (alb) <4 g/dL, hemoglobin (hem) <10.5 g/dL, stage IV, white blood cell (WBC) ≥15,000/mm³, lymphocyte (lymph) <600/mm³ and/or <8% of total WBC], Multivariable logistic regression was initially used to assess the association between bone marrow involvement (BMI) and each factor in the IPS. All 7 factors were significant (p<.001 for sex, age, albu, hem, stage and lymph; .07 for WBC); and therefore recursive partioning algorithm was used to identify a cutoff for determining BMI.

88 pts (10%) had BMI, we used a modified IPS formally by calculating a score that uses “weights” based on the results of a multivariable model including all 7 factors and the relative magnitude of the associated regression coefficients (RC), [modified IPS= WBC+2x(sex+age+hem)+3x(stage+alb)+4xlymph].

This model suggests lymphopenia has a stronger impact on the presence of BMI, while leukocytosis has the least impact. The percentage of BMI was 0.1,17,48, and 87 % for pts with a score of 0-8, 9-10, 11-12, and >12 respectively. Using a score of >8 as a rule for BMI, sensitivity was 99% and specificity was 87%. The prevalence of BMI by histology was 4% lymphocyte-rich, 5% nodular sclerosis, 20% mixed-cellularity and 21% lymphocyte-depleted, further analysis showed using modified IPS, sensitivity and specificity were similar across all types of histologies.

Using a modified IPS of > 8 for predicting BMI in HL doubled the spec associated with Ann Arbor classification. The implementation of modified IPS is a practical and reliable tool that will allow physicians to predict BMI in HL pts. Therefore eliminate unnecessary painful procedure in many pts.

No relevant conflicts of interest to declare.