Prognostic Value of CD38 Expression by Fluorescence Intensity Histogram in Chronic Lymphocytic Leukemia

Reviewer: Eric Shinohara MD, MSCI
Abramson Cancer Center of the University of Pennsylvania
Last Modified: March 31, 2007

Presenter: Barreto, W. G.
Presenter's Affiliation: Universidade Federal de São Paulo
Type of Session: Scientific


  • Chronic Lymphocytic Leukemia (CLL) is the most prevalent form of leukemia in the Western World.
  • CLL has a variable clinical course and CLL cells appear to proliferate at variable rates. 
  • Various different prognostic markers in CLL have been found including Beta2 microglobulin and Zap70.
  • CD38 has been found to be an independent prognostic marker in CLL however; the best CD38 cutoff to use remains controversial with values between 30% and 5% used previously.
  • Previous studies by Rassenti et al (2004) and Boonstra et al (2006) have found that CD38 expression is heterogenous in CLL cells. 
  • Flourescence intensity histogram (FIH) analysis to quantify CD38 expression may be useful as a prognostic marker in CLL.
  • This study was undertaken to classify patients into groups based on histograms of CD38 intensity and to correlate these grouping with prognosis.

Materials and Methods

  • This is a retrospective study of 51 patients with CLL accrued between 3/2002 and 8/2006.
  • Diagnosis of CLL was based on peripheral blood smear using WHO criteria.
  • The CD38 threshold level was based on CD28 histograms from 10 patients with CLL which was negative for CD38 expression.  The cursor was set at the right foot of this population, where <0.2% of cells were classified as positive.
  • Four patterns of FIH were described, negative (type I), weak expression (type II), strong expression (type III) and bimodal (type IV).
  • FIH’s association with Binet’s staging, CD38 positive percentage, and event-free survival (EFS) was studied.  EFS was defined as survival without treatment related to CLL.


  • 23.5% of patients were classified as CD38 negative, 39% as having weak expression, 14% as having strong expression and 23.5% demonstrated bimodal expression.
  • The mean percentage value of CD38, by type, was 2.1% for negative patients, 5.3% for weak expression, 57.3% for strong expression and 33.2% for bimodal expression.
  • EFS was 15, 14.4, 7.0, and 9.4 months for negative, weak expression, strong expression and bimodal expression, respectively. EFS in patients with bimodal expression was significantly different compared with patients with negative expression (p<0.05) and weak expression (p<0.05). 
  • 72% of patients with bimodal expression and 75% with strong expression required treatment at the time of diagnosis. 

There was a significant difference in EFS between FIH types in all Binet’s groups and this retained significance in Binet’s A patients alone, with bimodal patients having a lower EFS than any of the other types (p<0.05).

Author's Conclusions

  • Patients with a FIH demonstrating bimodal distribution had poor EFS.
  • Expression levels of CD38 as determined by FIH is a helpful tool in determining prognosis in CLL patients, particularly those with Binet’s stage A with CD38 expression <30%.

Clinical/Scientific Implications

  • This study addressed the need to find a sensitive yet specific threshold for the correlation between CD38 expression and prognosis. This study found that a bimodal FIH distribution was associated with a poor prognosis.
    • This study was retrospective and requires validation using the same FIH parameters in a prospective cohort of patients prior to its use as a prognostic marker.   
    • EFS may not be the best parameter to use as the implementation of CLL treatment may have some variability that could affect EFS validity, particularly in a small cohort of patients. A stricter definition, such as overall survival or specific parameters for progression free survival may be better suited for this.
    • This study found that between 72-75% of patients with bimodal and strong expression FIHs required treatment at diagnosis. However, EFS was still between 7-9.4 months in these patients, implying some had a long interval between diagnosis and treatment. Determining why these patients had a better prognosis would be of great interest.