Age at diagnosis and the utility of prognostic testing in patients with chronic lymphocytic leukemia

Authors


  • Presented at the American Society of Hematology Meeting, New Orleans, Louisiana, December 5-8, 2009.

Abstract

BACKGROUND:

A study was undertaken to analyze the survival of chronic lymphocytic leukemia (CLL) patients relative to age-matched individuals in the general population and determine the age-stratified utility of prognostic testing.

METHODS:

All 2487 patients diagnosed with CLL between January 1995 and June 2008 and cared for in the Mayo Clinic Division of Hematology were categorized by age at diagnosis and evaluated for differences in clinical characteristics, time to first treatment, and overall survival (OS).

RESULTS:

Among Rai stage 0 patients, survival was shorter than the age-matched general population for patients aged <55 years (P < .001), 55 to 64 years (P < .001), and 65 to 74 years (P < .001), but not those aged ≥75 years at diagnosis (P = not significant). CD38, IGHV mutation, and ZAP-70 each predicted time to first treatment independent of stage for all age groups (all P < .04), but had less value for predicting OS, particularly as age increased. IGHV and fluorescent in situ hybridization (FISH) predicted OS independent of stage for patients aged <55 years (P ≤ .001), 55 to 64 years (P ≤ .004), and 65 to 74 years (P ≤ .001), but not those aged ≥75 years. CD38 and ZAP-70 each predicted OS independent of stage for only 2 of 4 age categories. Among Rai 0 patients aged <75 years, survival was shorter than the age-matched population only for IGHV unmutated (P < .001) patients or those with unfavorable FISH (P < .001).

CONCLUSIONS:

Survival of CLL patients aged <75 years is shorter than the age-matched general population regardless of disease stage. Among patients aged <75 years, the simple combinations of stage and IGHV or stage and FISH identifies those with excess risk of death relative to the age-matched population. Although useful for predicting time to first treatment independent of stage for patients of all ages, prognostic testing had little utility for predicting OS independent of stage among patients aged ≥75 years. Cancer 2010. © 2010 American Cancer Society.

Chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL) is among the most common lymphoid malignancies, accounting for approximately 11% of hematologic cancers in the Western world.1 The prevalence of CLL increases with age, and the median age at the time of diagnosis is between 65 and 70 years.2-6 Recent studies suggest that the 5-year survival of CLL patients of all ages has increased over the past 2 decades,6-8 likely due in part to earlier stage at diagnosis.3, 4, 6, 9 The absolute 10-year survival of patients with CLL has increased by ∼10% for patients of all ages except those older than 80 years.7, 8

Although the observed improvement in the survival of CLL patients at the population level is encouraging, the clinical course of individual patients is heterogeneous. Even among individuals with early stage disease, there remains significant heterogeneity in clinical behavior, and stage alone does not adequately predict the risk of progression for a given patient.10 Although numerous clinical and biologic parameters are able to predict survival and time to first treatment,10-15 the utility of these prognostic parameters may vary based on age, given the higher mortality from competing health problems in older individuals.7, 16, 17 There remains a strong age gradient in the survival among CLL patients,2, 7, 8, 18 where the expected 10-year survival for those younger than 60 years is 59%, compared with 6% for those older than 80 years.7 Accordingly, although risk stratification using leukemia cell biomarkers (eg, ZAP-70, fluorescent in situ hybridization [FISH], IGHV testing) may provide useful information for counseling a newly diagnosed 50-year-old patient with Rai stage 0 disease, its usefulness to a 75-year-old patient in the same clinical circumstance is less clear.13, 16

These facts have important implications for the use of prognostic testing and counseling regarding life expectancy for older individuals with CLL who represent the majority of CLL patients worldwide. Most of the data on the ability of prognostic parameters to predict outcome is derived from cohorts of CLL patients with a median age <65 years,10-15 and the median age of patients in many series is <60 years.12, 14, 15 In the present study, we evaluated the clinical outcome of 2487 patients diagnosed with CLL between January 1995 and June 2008 to 1) evaluate differences in natural history based on age at diagnosis, 2) compare survival to age-matched individuals in the general population, and 3) determine the age-stratified utility of prognostic testing.

MATERIALS AND METHODS

Patients

The Mayo Clinic CLL Database includes all patients with a diagnosis of CLL19, 20 seen in the Division of Hematology at Mayo Clinic Rochester who permit their records to be used for research purposes.21-27 Clinical information regarding date of diagnosis, physical examination, clinical stage (Rai), prognostic parameters, treatment history, and disease-related complications are abstracted from clinical records on all patients at the time of inclusion and maintained on an ongoing, prospective basis. For staging purposes, patients with the small lymphocytic lymphoma variant of CLL who have cytopenias at diagnosis are grouped with Rai stage III/IV patients, whereas those with palpable lymphadenopathy without cytopenias are grouped with Rai stage I patients. Results of prognostic testing performed as part of clinical or research studies are also included in the database. This includes evaluation of absolute lymphocyte count (ALC), IGHV gene mutation analysis, ZAP-70 status, CD38 status, and cytogenetics abnormalities by interphase FISH testing using methods previously described by our group.21, 28-30

With the approval of the Mayo Clinic Institutional Review Board and in accord with federal regulations and the Declaration of Helsinki, we used this database to identify all patients diagnosed with CLL between January 1995 and June 2008. All these patients had an ALC ≥5.0 × 109/L and fulfilled the 1996 criteria for CLL, which were in effect throughout the study period,20 and/or fulfilled the World Health Organization criteria for the small lymphocytic lymphoma variant of CLL.19 Patients were categorized by age at the time of CLL diagnosis (<55 years, 55-64 years, 65-74 years, ≥75 years), with categories based in part on the previous designations of young CLL as individuals aged ≤55 years at diagnosis16, 31 and stratification of patients aged >55 years in 10-year intervals up to age 75 years. Differences in clinical characteristics, time to first treatment, and overall survival (OS) based on age at diagnosis were assessed. Because FISH can change during the course of the disease or after treatment,30 only FISH analysis obtained before first treatment was included in the present analysis. On the basis of evidence that VH3-21 family usage is associated with poor outcome independent of mutation status,32, 33 patients with VH3-21 family usage were considered to have high-risk IGHV status regardless of percentage mutation.

Statistical Analysis

OS was defined as the time between the date of diagnosis to the date of death or last follow-up. Time to first treatment was defined as the time between date of diagnosis and the date of initiation of first treatment or date of last follow-up at which the patient was known to be untreated. The accepted indications to initiate treatment were based on the National Cancer Institute Working Group 1996 criteria.20 Patients receiving early treatment as part of experimental protocols before meeting National Cancer Institute Working Group 1996 criteria to initiate therapy were censored as untreated on the date experimental therapy was administered. Estimates of survival were calculated using the Kaplan-Meier method. Cox proportional hazard models were used to model the relationship of multiple variables simultaneously, including age at diagnosis with OS and time to first treatment. Expected survival was calculated using the Cohort (Hakulinen) method34; estimates are based on the Minnesota White population.35 Likelihood ratio tests were used to test effects of individual factors either individually or jointly. Hazard ratios (HRs) and 95% confidence intervals were calculated from the Cox models. In addition to individual cytogenetic categories, FISH results were classified as unfavorable (17p−, 11q−) or favorable (normal, trisomy 12, 13q−, other) for some analyses. P values <.05 were considered significant. All statistical analyses were performed using the SAS 9.1 software package (SAS Institute; Cary, NC).

RESULTS

There were 2487 patients who qualified for inclusion in this study. The median age at diagnosis was 64 years. When grouped by age, 593 patients were aged <55 years, 713 aged 55 to 64 years, 748 aged 65 to 74 years, and 433 aged ≥75 years at diagnosis. The demographic and prognostic characteristics of patients by age at diagnosis are shown in Table 1. Because CD38, ZAP-70, IGHV gene mutation status, and FISH analysis were not routinely performed during the entirety of the study period, results were not available for all patients. Patients younger than 55 years were more likely to have intermediate stage (Rai I or II) at diagnosis, whereas those aged >65 years were less likely to have an ALC >30 × 109/L; however, no statistically significant differences were observed in CD38, IGHV mutation, ZAP-70, or the frequency of common cytogenetic abnormalities as identified by FISH based on age. Time to first treatment among Rai stage 0 patients was shorter for patients aged <55 years; however, no difference in time to first treatment by age was observed among Rai stage I or II patients.

Table 1. Patient Characteristics
CharacteristicAged <55 Years, n = 593Aged 55 to 64 Years, n = 713Aged 65 to 74 Years, n = 748Aged ≥75 Years, n = 433P
  • ALC indicates absolute lymphocyte count; FISH, fluorescent in situ hybridization; TFT, time to first treatment; NR, not reached; OS, overall survival.

  • a

    P value was not provided because the chi-square test was not appropriate, and calculation of the P value using the Fisher exact test was not feasible (too many FISH categories).

Median age at diagnosis, y49606980
Male404 (68%)493 (69%)497 (66%)295 (68%).739
Rai risk at diagnosis
 Low (Rai stage 0)251 (43%)383 (56%)408 (57%)245 (58%)<.001
 Intermediate (Rai stage I-II)308 (53%)257 (38%)269 (38%)129 (31%)
 High (Rai stage III-IV)21 (4%)40 (6%)40 (6%)46 (11%)
 Missing13333113 
ALC (×109/L)
 ≤30449 (77%)550 (79%)631 (85%)370 (86%)<.001
 >30133 (23%)148 (21%)109 (15%)60 (14%)
 Missing111583 
CD38
 Negative294 (66%)376 (69%)365 (68%)195 (65%).588
 Positive152 (34%)169 (31%)172 (32%)105 (35%)
 Missing147168211133 
ZAP-70
 Negative203 (61%)220 (60%)219 (66%)95 (66%).340
 Positive130 (39%)145 (40%)113 (34%)50 (34%)
 Missing260348416288 
IGHV
 Mutated150 (51%)172 (54%)170 (59%)64 (61%).166
 Unmutated144 (49%)144 (46%)119 (41%)41 (39%)
 Missing299397459328 
FISH (prior to treatment)
 13q−118 (44%)136 (43%)126 (41%)50 (34%)a
 Normal71 (27%)76 (24%)80 (26%)41 (28%)
 +1243 (16%)62 (20%)58 (19%)33 (23%)
 11q−22 (8%)26 (8%)23 (7%)16 (11%)
 17p−10 (4%)11 (3%)16 (5%)6 (4%)
 Other3 (1%)5 (2%)4 (1%)1 (1%)
 Missing326397441286 
FISH groupings
 Normal, 13q−, trisomy 12232 (88%)274 (88%)264 (87%)124 (85%).799
 11q−, 17p−32 (12%)37 (12%)39 (13%)22 (15%)
Treated268 (45%)282 (40%)296 (40%)108 (25%)<.001
 Median TFT, Rai 0 patients5.98.88.6NR.047
 Median TFT, Rai I-II patients1.92.52.53.4.216
Dead112 (19%)171 (24%)254 (34%)199 (46%)<.001
 Median OS, y11.810.99.06.4<.001

Median follow-up was 9.7 years. As of last follow-up, 954 patients had been treated, and 727 patients had died. Median time to first treatment was 4.8 years. Although survival decreased as the age at diagnosis increased (Fig. 1A), the survival of CLL patients was significantly shorter than that of the age-matched general population for patients aged <55 years (P < .001), 55 years to 64 years (P < .001), and 65 to 74 years (P < .001) at diagnosis, but not those aged ≥75 years (P = .14; Fig. 1B-E). Among Rai stage 0 patients, survival was also shorter than that of the age-matched general population for CLL patients aged <55 years (P < .001), 55 to 64 years (P < .001), and 65 to 74 years (P < .001), but not those aged ≥75 years (P = .07) at diagnosis (Fig. 2).

Figure 1.

Survival of patients with chronic lymphocytic leukemia (CLL) compared with age-matched individuals is shown. (A) Survival based on age at diagnosis is shown (age <55 years [n = 593], ages 55-64 years [n = 713], ages 65-74 years [n = 748], and age ≥75 years [n = 433]). (B) Survival is shown for patients aged <55 years at the time of diagnosis (n = 593) compared with the age-matched population aged <55 years. (C) The survival of CLL patients aged 55 to 64 years at the time of diagnosis (n = 713) is shown compared with the age-matched population aged 55 to 64 years. (D) The survival of CLL patients aged 65 to 74 years at the time of diagnosis (n = 748) is shown compared with the age-matched population aged 65 to 74 years. (E) The survival of CLL patients aged >75 years at the time of diagnosis (n = 433) is shown compared with the age-matched population aged >75 years.

Figure 2.

Survival of patients with Rai stage 0 chronic lymphocytic leukemia (CLL) is shown compared with age-matched individuals. (A) The survival of patients with Rai stage 0 CLL who were aged <55 years at the time of diagnosis (n = 248) is shown compared with the age-matched population aged <55 years. (B) The survival of patients with Rai stage 0 CLL who were aged 55 to 64 years at the time of diagnosis (n = 379) is shown compared with the age-matched population aged 55 to 64 years. (C) The survival of patients with Rai stage 0 CLL who were aged 65 to 74 years at the time of diagnosis (n = 405) is shown compared with the age-matched population aged 65 to 74 years. (D) The survival of patients with Rai stage 0 CLL who were aged >75 years at the time of diagnosis (n = 244) is shown compared with the age-matched population aged >75 years.

We next evaluated the relationship between prognostic parameters and time to first treatment and OS. Consistent with prior reports, stage, ALC (≤ or >30 × 109/L), CD38, IGHV mutation, ZAP-70, and cytogenetic analysis by FISH were all powerful predictors of both time to first treatment and OS on univariate analysis (all P ≤ .003 for both time to first treatment and OS; Table 2). In a multivariate analysis in the 585 patients who had results for all prognostic variables, stage (HR for high Rai risk vs low Rai risk, 16.1; HR for intermediate Rai risk vs low Rai risk, 2.3), IGHV unmutated (HR, 2.8; P < .001), ALC >30 × 109/L (HR, 2.0), and CD38 positive (HR, 1.9) remained independent predictors of time to first treatment (all P < .001), whereas ZAP-70 (HR, 1.3; P = .16) and FISH (HR, 1.3; P = .264) were no longer statistically significant. With respect to OS, only high-risk FISH (HR, 2.9; P = .008) and CD38 positive (HR, 2.2; P = .041) remained independent predictors of OS, whereas IGHV (HR, 2.8; P = .072), stage (HR for high Rai risk, 2.8 [P = .20]; HR for intermediate Rai risk, 1.4 [P = .37]), ALC (HR, 1.2; P = .62), and ZAP-70 (HR, 0.9; P = .77) were no longer statistically significant.

Table 2. Prognostic Parameters, TFT, and OS
ParameterMedian TFT, YearsPMedian OS, YearsP
  1. TFT indicates time to first treatment; OS, overall survival; ALC, absolute lymphocyte count; NR, not reached; FISH, fluorescent in situ hybridization.

Rai risk at diagnosis (n = 2397)
 Low (Rai stage 0)8.0<.00111.4<.001
 Intermediate (Rai stage I-II)2.48.8
 High (Rai stage III-IV)0.15.2
ALC (×109/L) (n = 2450)
 ≤305.6<.00110.2.003
 >302.4 8.7
CD38 (n = 1828)
 Negative8.0<.00111.9<.001
 Positive2.88.5
ZAP-70 (n = 1175)
 Negative9.3<.001NR<.001
 Positive3.210.8
IgVH Mutation (n = 1004)
 Mutated11.0<.001NR<.001
 Unmutated2.89.7
FISH (prior to treatment) (n = 715)
 13q−NR<.001NR<.001
 Normal8.7NR
 +125.410.9
 11q−2.48.4
 17p−5.27.6
FISH groupings
 Normal, 13q−, trisomy 128.7<.001NR<.001
 11q−, 17p−3.77.6

Given the variation in time to first treatment and OS based on age at diagnosis as well as differences in the magnitude of effect of a CLL diagnosis on survival relative to age-matched controls, we next evaluated whether ALC, CD38, IGHV mutation, ZAP-70, and FISH remained useful predictors of time to first treatment and OS for CLL patients in all age categories. Stage, ALC, CD38, ZAP-70, and IGHV remained powerful predictors of time to first treatment for CLL patients of all ages including those aged ≥75 years (all P ≤ .005). FISH was also a powerful predictor of time to first treatment for CLL patients of all ages except those older than 75 years, where it failed to reach statistical significance (P = .09). The hazard ratio of each individual prognostic parameter for predicting time to first treatment was generally similar across age groups.

In contrast to the near uniform value of ALC, CD38, IGHV mutation, ZAP-70, and FISH for predicting time to first treatment in all age categories, their ability to predict OS was less consistent. As univariate factors, stage, CD38, and IGHV were statistically significant predictors of OS in CLL patients of all age categories, including those aged ≥75 years (all P < .05). FISH predicted OS in all age categories <75 years (all P ≤ .003), but not in patients aged ≥75 years (P = .34). ZAP-70 predicted OS for patients who were aged <55 years (P = .007) and 55 to 64 years (P = .004), but not patients aged 65 to 74 years (P = .28) or >75 years (P = .97). ALC >30 × 109/L was only a significant predictor of OS for patients in the 65- to 74-year-old group.

Next, we evaluated the ability of ALC, CD38, IGHV mutation, ZAP-70, and FISH to predict time to first treatment and OS in each age category after adjusting for stage. With respect to time to first treatment, CD38, IGHV mutation, and ZAP-70 each predicted time to first treatment independent of stage for all age categories (Table 3, Top). FISH predicted time to first treatment independent of stage for patients aged <55, 55 to 64, and 65 to 74 years, but not for those aged ≥75 years (P = .08). The utility of these parameters for predicting OS independent of stage was less consistent and varied by age. Both IGHV and FISH predicted OS independent of stage for patients aged <55, 55 to 64, and 65 to 74 years but not those aged ≥75 years (Table 3, Bottom). CD38 and ZAP-70 each predicted survival independent of stage for only 2 of the 4 age categories (ZAP-70 for age <55 and 55 to 64 years; CD38 for 55 to 64 and ≥75 years).

Table 3. Multivariate Models
Time to First Treatment
Prognostic FactorAged <55 YearsAged 55 to 64 YearsAged 65 to 74 YearsAged ≥75 Years
HRPHRPHRPHRP
  • HR indicates hazard ratio; ALC, absolute lymphocyte count; UM, unmutated; M, mutated, FISH, fluorescent in situ hybridization.

  • a

    Too few events to estimate.

Rai stage: high vs low7.7<.0019.0<.0016.7<.00110.6<.001
Rai stage: intermediate vs low2.5<.0012.4<.0012.8<.0014.3<.001
ALC (>30 vs ≤30)1.3.0701.8<.0012.2<.0012.9<.001
Rai stage: high vs low14.6<.00116.0<.0016.9<.00111.3<.001
Rai stage: intermediate vs low2.4<.0012.4<.0012.6<.0012.6<.001
CD38 (positive vs negative)2.3<.0012.8<.0011.7<.0011.7.037
Rai stage: high vs low19.7<.0019.7<.00115.0<.0016.8.001
Rai stage: intermediate vs low2.3<.0012.2<.0013.6<.0011.1.802
ZAP-70 (positive vs negative)2.0<.0011.9<.0012.8<.0015.6<.001
Rai stage: high vs low16.1<.0018.6<.00130.0<.00111.7<.001
Rai stage: intermediate vs low2.7<.0012.1<.0012.5<.0011.3.587
IgVH mutation (UM vs M)3.8<.0013.6<.0014.7<.0015.4.002
Rai stage: high vs low6.2<.0016.2<.00110.0<.00113.2<.001
Rai stage: intermediate vs low2.8<.0012.1.0024.9<.0013.2.003
FISH (17p−, 11q− vs normal, 13q−, trisomy 12)2.7.0013.0<.0012.8<.0012.0.078
Overall Survival
Prognostic FactorAged <55 YearsAged 55 to 64 YearsAged 65 to 74 YearsAged ≥75 Years
HRPHRPHRPHRP
Rai stage: high vs low3.6.0025.0<.0012.6<.0013.8<.001
Rai stage: intermediate vs low2.2<.0011.9<.0011.9<.0011.8<.001
ALC1.4.1261.5.0271.7.0011.5.052
Rai stage: high vs low11.8<.0014.4<.0012.1.0473.2<.001
Rai stage: intermediate vs low2.3.0081.7.0211.9.0012.0.003
CD38 (positive vs negative)1.6.0772.2<.0011.3.1381.5.038
Rai stage: high vs low22.8.0059.1<.0011.0.9829.1<.001
Rai stage: intermediate vs low2.8.0131.4.3332.1.0244.1<.001
ZAP-70 (positive vs negative)2.6.0112.2.0251.2.5780.7.426
Rai stage: high vs low34.0<.0011.7.4752.1.3192.7.358
Rai stage: intermediate vs low2.2.0741.5.2252.0.0191.8.120
IgVH mutation (UM vs M)6.2<.0012.8.0043.2<.0012.0.073
Rai stage: high vs lowa 4.5.0077.4<.0013.7.002
Rai stage: intermediate vs low1.7.2941.0.9362.3.0072.7.008
FISH (17p−, 11q− vs normal, 13q−, trisomy 12)5.2.0014.6.0013.0.0011.4.372

Finally, because OS was only shorter than the age-matched population for CLL patients age <75 years, we evaluated whether prognostic testing could identify which Rai 0 CLL patients aged <75 years had a survival shorter than the age-matched population. IGHV and FISH testing were used for this analysis based on the ability of these tests to identify CLL patients with shorter OS independent of stage in all age categories <75 years. The survival of Rai 0 IGHV unmutated CLL patients aged <75 years was shorter than that of the age-matched general population (P < .001), whereas Rai 0 IGHV mutated CLL patients had a survival similar to the population (Fig. 3A and B). Similarly, the survival of Rai 0 CLL patients aged <75 years with unfavorable FISH was shorter than that of the age-matched general population (P < .001), whereas Rai 0 CLL patients with favorable FISH had survival similar to the age-matched general population (Fig. 3C and D).

Figure 3.

Survival of patients with Rai stage 0 chronic lymphocytic leukemia (CLL) aged <75 years is shown compared with age-matched individuals based on IGHV mutation status and cytogenetic analysis by fluorescent in situ hybridization (FISH). (A) Survival of patients with Rai stage 0 CLL and mutated IGHV aged <75 years (n = 344) is shown compared with the age-matched population. (B) Survival of patients with Rai stage 0 CLL with unmutated IGHV aged <75 years (n = 160) is shown compared with the age-matched population. (C) Survival of patients with Rai stage 0 CLL aged <75 years (n = 317) with favorable FISH results (no 17p− or 11q−) is shown compared with the age-matched population. (D) Survival of patients with Rai stage 0 CLL aged <75 years (n = 27) with unfavorable FISH results (eg, 17p− or 11q−) is shown compared with the age-matched population.

DISCUSSION

Age has repeatedly been shown to be an independent predictor of survival in CLL patients,16, 17, 36-38 and has the potential to alter the utility of prognostic testing given the higher mortality from competing health problems in older individuals. Although the majority of patients with CLL are older than 65 years at the time of diagnosis and have early stage disease,3, 4, 6, 9 most of the published data on prognostic parameters is derived from younger patient cohorts and often include advanced stage patients. This incongruity has led to uncertainty regarding whether and when to use prognostic testing in routine clinical practice for patients with CLL. In the present study, the survival of CLL patients aged <75 years at diagnosis was shorter than that of the age-matched general population regardless of disease stage. In contrast, survival did not differ from the age-matched general population among CLL patients aged ≥75 years at diagnosis. Prognostic testing using CD38, IGHV mutation, and ZAP-70 was useful for predicting time to first treatment independent of stage for CLL patients of all ages (including those aged ≥75 years), but had less value for predicting OS, particularly as the age at diagnosis increased. Among Rai 0 patients aged <75 years, survival was shorter than the age-matched general population only for IGHV unmutated patients or those with unfavorable FISH.

These findings have several important implications for the use of prognostic tests in patients with CLL as well as the use of test results to select patients for clinical trials testing the value of early treatment. First, although the life expectancy of CLL patients aged <75 years is substantially shorter than that of the age-matched general population, the simple combination of stage and IGHV mutation status or stage and FISH can identify those with excess risk of death. Second, with regard to time to first treatment, CD38, ZAP-70, IGHV, and FISH each provide useful information for CLL patients of all ages independent of stage. Third, prognostic markers appear to be a sound basis on which to select high-risk early stage patients aged <75 years for clinical trials of early intervention. In the present cohort, both IGHV mutation status and FISH predicted OS independent of stage, with relatively large hazard ratios (range, 2.8-6.2) among patients in all age categories <75 years. Fourth, although useful for predicting time to first treatment among patients aged ≥75 years, prognostic testing appears to have more limited utility for predicting OS independent of stage among patients in this age category. Accordingly, it does not appear appropriate to enroll patients aged >75 years in clinical trials of early intervention based on prognostic testing if the aim is to improve OS.

Several other aspects of this analysis are noteworthy. The study included comprehensive multivariate analysis of both traditional (age, stage, ALC) and biologic (ZAP-70, CD38, FISH, IGHV) parameters in a large cohort of CLL patients. Consistent with prior reports demonstrating that the biologic parameters contain complementary prognostic information,14, 39, 40 multiple biologic prognostic parameters were independent predictors of time to first treatment and/or OS in multivariate analyses including traditional parameters (stage, ALC). It is also notable that the distribution of biologic prognostic parameter results did not differ by age at diagnosis, arguing against the notion that CLL in younger patients is more biologically aggressive.

How do these results relate to previous studies? Although age has repeatedly been shown to be an independent predictor of survival in CLL patients,7, 8, 16, 17 few prior studies have evaluated interactions between age and the utility of prognostic testing. Mauro and colleagues previously demonstrated that lymphocyte doubling time is a predictor of OS among CLL patients both aged ≤55 years and aged >55 years, but provided no further age stratification of those older than 55 years.31 Dohner and colleagues observed that the presence of del(11q22) was a profound stratifier of survival in patients aged <55 years (median survival, del[11q22] = 64 months vs no del[11q22] = 209 months; P < .001) but not patients aged >55 years (median survival, del[11q22] = 94 months vs no del[11q22] = 111 months; P = .82).16 Because of the limited data demonstrating utility in older patients, many hematologist/oncologists do not routinely use prognostic testing for older individuals with early stage CLL, which limits the accuracy of counseling on natural history and life expectancy for these patients. The present study provides more comprehensive data regarding the utility of prognostic testing for classifying risk among CLL patients older than 65 years. To our knowledge, it also is the first study to evaluate how prognostic test results can be used to stratify the risk of death among patients with CLL relative to the age-matched general population.

Our study has several important limitations. Given that the natural history of CLL is changing,7, 8 the fact that all patients in the current cohort were diagnosed in the past 15 years is a strength of the study. However, not all patients had all of the molecular/biologic prognostic parameters measured since they were only discovered/used routinely in the past 5 to 10 years.10-14 Second, the diagnosis of CLL was based on the 1996 criteria for CLL,20 which were in effect throughout the study interval but recently underwent revision.41 Third, as in other analyses of clinical outcome,36, 42-44 we evaluated OS rather than disease-specific survival. Because OS is the outcome of greatest interest to patients, we believe that this is the most appropriate outcome for survival analysis, with use of time to first treatment as a secondary measure of disease-specific outcomes. Fourth, our study focused on the ability of prognostic tests to predict time to first treatment and OS. The utility of these assays for predicting other outcomes, such as response to treatment, was not the focus of the current study. Finally, the study represents a single-center experience that requires validation in independent series of patients monitored prospectively.

In aggregate, these findings suggest that survival of CLL patients aged <75 years at diagnosis is shorter than that of the age-matched general population regardless of disease stage. Prognostic testing had little utility for predicting OS independent of stage among patients aged ≥75 years, although it remained useful for predicting time to first treatment. In settings where the goal is to identify patients with excess risk of death relative to the age-matched population, it appears that clinical staging is the only test necessary for patients aged ≥75 years at diagnosis, whereas limited testing with a combination of stage and IGHV or stage and FISH are appropriate strategies for most patients aged <75 years.

CONFLICT OF INTEREST DISCLOSURES

Support through grants from the National Cancer Institute (CA 113,408 to T.D.S. and CA136591 to D.F.J.) and Gabrielle's Angel Foundation for Cancer Research (T.D.S.) are gratefully acknowledged.

Ancillary