Conflict of interest: Nothing to report
Article first published online: 1 AUG 2012
Copyright © 2012 Wiley Periodicals, Inc.
American Journal of Hematology
Volume 87, Issue 11, pages 1006–1009, November 2012
How to Cite
Komrokji, R. S., Corrales-Yepez, M., Kharfan-Dabaja, M. A., Al Ali, N. H., Padron, E., Rollison, D. E., Pinilla-Ibarz, J., Zhang, L., Epling-Burnette, P. K., Lancet, J. E. and List, A. F. (2012), Hypoalbuminemia is an independent prognostic factor for overall survival in myelodysplastic syndromes. Am. J. Hematol., 87: 1006–1009. doi: 10.1002/ajh.23303
Data were Presented at the American Society of Hematology 52nd Annual Meeting, December 2010, Orlando, Florida (Abstract # 1022).
- Issue published online: 23 OCT 2012
- Article first published online: 1 AUG 2012
- Accepted manuscript online: 28 JUN 2012 07:52AM EST
- Manuscript Accepted: 21 JUN 2012
- Manuscript Revised: 20 JUN 2012
- Manuscript Received: 18 MAY 2012
We hypothesized that hypoalbuminemia is an independent prognostic factor in patients with myelodysplastic syndromes (MDS). We analyzed records of 767 patients treated at Moffitt Cancer Center between January 2001 and December 2009 to evaluate the relationship between serum albumin (SA) at the time of presentation and overall survival (OS). Patients (median age of 69 years) were stratified into three groups based on SA concentration (≤3.5, 3.6–4.0, and >4.0 g/dL). Two-thirds of the patients had low or intermediate-1 International Prognostic Scoring System (IPSS)-based risk for MDS. Median OS by SA concentration of ≤3.5, 3.6–4.0, and >4.0 g/dL was 11, 23, and 34 months, respectively (P < 0.005), whereas rate of acute myeloid leukemia progression was highest in patients with low SA (≤3.5 g/dL). The SA level offered prognostic discrimination for outcomes within the lower and higher IPSS risk groups, as well as with the MD Anderson risk model. In multivariable analysis, SA was a significant independent co-variate for OS after adjustment for IPSS, age, serum ferritin, and transfusion dependence (hazard ratio = 0.8; 95% CI 0.6–0.9; P = 0.004). Our findings indicate that hypoalbuminemia is an independent prognostic biomarker that may serve as a surrogate representative of disease biology or comorbidities in patients with MDS. Am. J. Hematol., 2012. © 2012 Wiley Periodicals, Inc.
Myelodysplastic syndromes (MDS) are a spectrum of hematopoietic neoplasms characterized by bone marrow failure and predisposition to acute myeloid leukemia (AML) progression [1, 2]. The hallmark of early disease is accelerated apoptosis due to intrinsic susceptibility of the clone, cytokine release, and immunological de-arrangements . The treatment for MDS is tailored based on risk stratification utilizing different clinical predictive models; the most widely used model is the International Prognostic Scoring System (IPSS)  with other factors such as comorbidities being very important to incorporate in treatment decisions .
A low-serum albumin level is a recognized adverse prognostic factor in patients with malignancies such as multiple myeloma, melanoma, and colon cancer [6–8]. We previously reported that severe hypoalbuminemia (<3.0 g/dL) at day +90 post-allogeneic hematopoietic stem cell transplant is an independent predictor of non-relapse and overall mortality in patients with AML and MDS . We also reported a similar prognostic value of serum albumin level in patients with relapsed/refractory AML .
As a biomarker affected by the disease process and comorbidities, we hypothesize that hypoalbuminemia may be an objective prognostic co-variate for outcome in patients with MDS. In this study, we examined the prognostic value of serum albumin levels in patients with MDS and its interaction with the IPSS category and other prognostic features.
Data were analyzed from the Moffitt Cancer Center (MCC) MDS database supplemented by individual chart review. The primary objective was to examine the role of serum albumin at time of presentation to MCC as a prognostic marker for overall survival (OS) and AML progression, as well as to assess its interaction with other prognostic variables such as IPSS, serum ferritin, red blood cell transfusion dependence (RBC-TD), and age. Assessment of the IPSS prognostic score was performed as described earlier . Patients were stratified into three groups based on serum albumin concentration: ≤3.5, 3.6–4.0, and >4.0 g/dL. The OS was calculated from time of referral to MCC. The Kaplan–Meier method was used to estimate median OS. The log-rank test was used to compare Kaplan–Meier survival estimates between two groups. Cox proportional hazards regression was used for multivariable analysis.
Between January 2001 and December 2009, 767 patients were captured in the MCC MDS database. Median age was 69 years, and the median duration of follow up from time of referral to MCC was 35 months. MDS subtypes were coded as refractory anemia (22%; n = 166), refractory anemia with ring sideroblasts (RARS; 12%; n = 93), del(5q) (3%; n = 23), refractory cytopenia with multi-lineage dysplasia (19%; n = 143), refractory anemia with excess blasts (42%; n = 324), and MDS unclassified (2%; n = 18). Distribution of IPSS risk groups was as follows: 19% low risk, 44% intermediate-1 (int-1), 21% int-2, 5% high-risk, and 11% unknown. The mean serum albumin was 3.87 g/Dl (SD 0.47 g/dL). Baseline characteristics for the three patient groups defined by serum albumin level are summarized in Table I. Patients with low serum albumin (≤3.5 g/dL) had more high-risk IPSS (P = 0.01) and higher mean serum ferritin (P = 0.001). There were no differences in the rate of RBC-TD among the three groups (P = 0.7) and no differences in proportion of patients who received azanucleosides (P = 0.3). Median OS for all patients was 22 months (95% CI 19.0–26). Age, IPSS risk group, RBC-TD, and serum ferritin were statistically significant prognostic variables in univariable analysis (Table II). As shown in Fig. 1, median OS results by level of serum albumin were as follows: 11 months (95% CI 7–14 months), 23 months (95% CI 19–27), and 34 months (95% CI 25–42) for patients with serum albumin levels of ≤3.5, 3.6–4.0, and >4.0 g/dL, respectively (P < 0.005, log-rank test).
|Albumin ≤3.5g/dL (n = 147)||Albumin 3.6–4.0 g/dL (n = 322)||Albumin >4.0 g/dL (n = 273)|
|Mean age, years||68||68||65|
|IPSS, no. of patients (%)||n = 131||n = 291||n = 252|
|Low||15 (12%)||68 (23%)||61 (24%)|
|Int-1||63 (48%)||140 (48%)||126 (50%)|
|Int-2||37 (28%)||69 (24%)||54 (21%)|
|High||16 (12%)||14 (5%)||11 (5%)|
|Mean serum ferritin level (ng/mL)||2,263||1,070||809|
|RBC-TD, No. of patients (%)||74 (51%)||157 (49%)||121 (45%)|
|Azanucleosides treatment, no. of patients (%)||65 (44%)||158 (49%)||142 (51%)|
|Median OS (months)||95% CI||P value|
As shown in Fig. 2, among patients in the low/int-1 IPSS risk group, median OS was 17 months (95% CI 11–23), 28 months (95% CI 23–33), and 41 months (95% CI 27–55) for patients with serum albumin levels of ≤3.5, 3.6–4.0, and >4.0 g/dL, respectively (P = 0.001). Among patients in the int-2/high IPSS risk group, median OS was 6 months (95% CI 3–10), 12 months (95% CI 9–16), and 15 months (95% CI 12.0–19.8), respectively, for patients with serum albumin levels of ≤3.5, 3.6–4.0, and >4.0 g/dL, respectively (P < 0.005). The same observation was noted among patients risk stratified as low/int-1 group by the global MD Anderson Risk Model (MDAS), where median OS was 25 months (95% CI 12–39), 34 months (95% CI 24–45), and 52 months (95% CI 32–72) for patients with serum albumin levels of ≤3.5, 3.6–4.0, and >4.0 g/dL, respectively (P = 0.03). In int-2/high-risk MDAS groups, the median OS was 6 months (95% CI 3–9), 12 months (95% CI 10–14), and 16 months (95% CI 14–19) for patients with serum albumin levels of ≤3.5 g/dL, 3.6–4.0 g/dL, and >4.0 g/dL, respectively (P < 0.005).
Hypoalbuminemia was predictive for OS in patients who were less than 60 years old, with median OS of 14 months (95% CI 10–18), 32 months (95% CI 22–42), and 37 months (95% CI 31–42), respectively, for patients with serum albumin levels of ≤3.5, 3.6–4.0, and >4.0 g/dL, respectively (P = 0.005). The same relation was found for patients who were ≥60 years old, with median OS of 10 months (95% CI 6–13), 20 months (95% CI 15–24), and 31 months (95% CI 20–42), respectively (P < 0.005).
In multivariable analysis using Cox regression, serum albumin was a statistically significant independent co-variate after adjustment for IPSS, age >60 years, and RBC-TD and serum ferritin. OS was superior among MDS patients with higher serum albumin (hazard ratio = 0.8; 95% CI 0.6–0.9; P = 0.004; Table III). Serum albumin was also statistically significant independent factor for OS after adjustment for MDAS.
|Hazard ratio||95% CI||P value|
|Age >60 years||1.3||0.96–1.8||0.08|
|Serum ferritin >1,000 ng/mL||1.4||1.1–1.9||0.006|
The overall rate of AML transformation was 28% (n = 212), 67% (n = 514) did not have documented AML transformation, and information on AML transformation were missing in 5% (n = 41). Rate of AML progression was highest in patients with lower serum albumin (40% in patients with serum albumin levels of ≤3.5 g/dL versus 28% in patients with levels of 3.6–4.0 g/dL and 22% in patients with levels >4.0 g/dL) (P = 0.006).
Several prognostic models have been developed to estimate OS and progression to AML in MDS [4, 11, 12]. IPSS is the most widely utilized prognostic tool for tailoring treatment selection according to expected disease behavior . New risk models, including the World Health Organization-based prognostic scoring system and the MDAS, have been recently validated as dynamic tools refining the IPSS [11, 12]. Those models depend mostly on intrinsic disease features such as percentage of bone marrow myeloblasts, cytogenetic abnormalities, number and depth of cytopenias, and RBC-TD. Other variables such as serum ferritin, serum lactate dehydrogenase, beta-2 microglobulin, and bone marrow fibrosis have been reported as independent prognostic variables [13–16]. Of late, genomic profiling and advanced molecular testing have shown that other important biological prognostic factors such as gene mutations (TET-2, ASXL-1, and C-cbl) offer further prognostic discrimination; however, these have not as yet been incorporated into routine clinical practice [17–19].
To our knowledge, this is the first study to demonstrate that serum albumin is an independent prognostic factor for OS in MDS patients in a large cohort. We were able to segregate patients into three prognostic groups based on serum albumin concentration (≤3.5, 3.6–4.0, and >4.0 g/dL) with distinct OS and rates of AML transformation. The prognostic value of low serum albumin was greatest among patients in the low/int-1 IPSS risk group but remained an independent prognostic variable across all risk groups. Hypoalbuminemia remained an independent prognostic factor in multivariable analysis after adjusting for age, IPSS, and RBC-TD and was also independent from MDAS.
Serum albumin has been reported as a prognostic variable in many medical conditions such as end-stage renal disease , coronary artery disease, and postoperative surgical outcomes . It is a surrogate marker of general health, comorbidities, nutrition, and performance status. Among malignancies, hypoalbuminemia is associated with an inferior outcome in colon cancer, melanoma, and multiple myeloma [6–8]. Kharfan-Dabaja et al. reported that hypoalbuminemia (serum albumin level of <3.0 g/dL) 90 days after allogeneic stem cell transplantation was associated with a significantly higher non-relapse mortality and reduced OS for patients with AML and MDS. The estimated 1- and 2-year cumulative incidence of non-relapse mortality and OS rates were 48 and 7% and 52 and 3%, respectively, for patients with serum albumin of <3.0 g/dL compared to ≥3.0 g/dL .
In multiple myeloma, serum albumin has been incorporated into the International Prognostic Index for disease staging . Interleukin-6 plays an important role in myeloma cell proliferation, survival, and resistance to treatment . Tumor necrosis factor-α, interleukin-1, and interleukin-6 are soluble inflammatory cytokines that directly suppress albumin synthesis . Inflammatory cytokines are thought to play an important role in the biology of MDS. High levels of tumor necrosis factor-α and interleukin-6 are characteristic of the circulating cytokine profile in patients with MDS , raising the prospect that lower serum albumin levels in patients with MDS could arise as a result of the actions of these inflammatory cytokine effectors. Interestingly, lenalidomide and azacitidine, agents with the potential to improve hematopoiesis in MDS, modify the cytokine milieu. Other factors such as nutritional status and comorbidities, the latter of which is recognized as an independent prognostic modifier in MDS , would be expected to influence serum albumin concentration, and, as such, serum albumin might serve as a surrogate biomarker for comorbidity score. Further investigation is needed to determine whether albumin could effectively replace comorbidity score for the purposes of prognostic assessment.
The limitations of our study include the retrospective study design, lack of detailed comorbidities history or comorbidity scores. The data on AML transformation were as of last follow up at our center. Our database lack information regarding cytokine levels, and finally the fact that serum albumin can be affected by many other variables as an acute phase reactant. Prospective validation of hypoalbuminemia as a prognostic biomarker is needed to confirm its utility.
We thank Rasa Hamilton (Moffitt Cancer Center) for editorial assistance.
- 1The myelodysplastic syndromes. In: Abeloff MD, Armitage JO, Niederhuber JE, editors. Clincial Oncology. New York: Churchill Livingstone; 2004. pp 2849–2881., , .
- 10Albumin is a prognostic factor for response and overall survival in relapsed or refractory acute myeloid leukemia (AML). ASH Annual Meeting Abstracts 2009; 114: 4685., , , et al.
- 15Independent impact of iron overload and transfusion dependency on survival and leukemic evolution in patients with myelodysplastic syndrome. ASH Annual Meeting Abstracts 2008; 112: 640., , , et al.