The effect of immunoglobulin VH gene mutation status and other prognostic factors on the incidence of major infections in patients with chronic lymphocytic leukemia

Authors


Abstract

BACKGROUND.

Infections are a major factor in the clinical course of chronic lymphocytic leukemia (CLL) and account for 30% to 50% of all deaths. The pathogenesis of infections in CLL is related to hypo-γ-globulinemia, T-cell immune dysfunction, and the immunosuppressive effect of treatment.

METHODS.

The authors retrospectively assessed the correlations between new prognostic markers and types of infections encountered, the time taken to develop these infections, and infection-related mortality in 280 unselected patients with CLL.

RESULTS.

One hundred patients (36%) had at least 1 major infection (median, 2 major infections; range, 1–8 major infections) over a median follow-up of 67 months. Infections were the most common cause of death, accounting for 51% of all fatalities. Older age (P = .007), clinical Stage B or C disease (P < .001), unmutated immunoglobulin (Ig)VH gene status (P < .001), genetic abnormalities (P < .001), positive CD38 status (P < .001), and type of initial therapy were associated with a significantly shorter time to first infection. Equally, patient age (P < .001), disease stage (P < .001), CD38 expression (P < .001), IgVH mutation status (P < .001), and genetic abnormalities (P = .003) had a significant impact on infection-related mortality.

CONCLUSIONS.

Clinical stage at diagnosis, IgVH mutation status, and initial therapy were possible predictors of severe infections in patients with CLL. The current results may help to identify which patients with CLL are at particularly high risk of developing serious infections and, thus, should be considered for Ig or antibiotic prophylaxis. Cancer 2006. © 2006 American Cancer Society.

Chronic lymphocytic leukemia (CLL) is a disorder characterized by increased numbers of small, mature lymphocytes in the blood and bone marrow. The disease exhibits clinical heterogeneity, and the survival of patients with CLL ranges considerably from 1 month to >20 years.1 This is related to differences in the clinical response to chemotherapy, the rate of disease progression, and the clinical stage at diagnosis. Initially, patients can present with limited disease that does not require immediate treatment, but a significant proportion will progress and require treatment. New prognostic markers, such as CD38 expression, immunoglobulin (Ig)VH mutation status, ZAP-70 expression, and cytogenetic abnormalities, have aided in stratifying patients' risk.2 High levels of CD38 and ZAP-70 expression3, 4 and unmutated IgVH genes3, 5 are associated with aggressive disease. Trisomy 12, complex karyotype, and deletions of chromosomic regions 11q and 17p are associated with reduced overall survival.6

Infections are a major factor in the clinical course of CLL. Up to 50% of patients with CLL will develop infectious complications during their disease course, and infections still account for 30% to 50% of all deaths from CLL.7 The pathogenesis of infections in CLL is related to a combination of hypo-γ-globulinemia, T-cell deficiency and the immunosuppressive effect of treatment.7, 8 Hypo-γ-globulinemia caused by a progressive decline in B-cell numbers and function is correlated with recurrent severe bacterial infections. Decreased IgG and IgA levels have been associated with reduced survival.9 Neutropenia from progressive bone marrow infiltration and chemotherapy also predispose patients to bacterial and fungal infections. T-cell dysfunction and depletion as a result of suboptimal antigen presentation and inhibition of T-cell responses by cytokines secreted by the malignant cells also can lead to opportunistic infections.10 Recent advances in treatment for CLL have altered the pattern and type of infections encountered. It was demonstrated that fludarabine and alemtuzumab improved response rates compared with traditional alkylating agents but also caused significant immunosuppression.11, 12 Autologous and allogeneic hematopoietic cell transplantation (HCT) also are associated with prolonged neutropenia and immunosuppression, which results in fungal, viral, and other opportunistic infections.13, 14

The objective of the current study was to assess the relations between new prognostic markers and therapies on the type of infections encountered, the time taken to develop these infections, and the infection-related mortality (IRM) rate. For this study, we also attempted to identify which patients with CLL were at particularly high risk of developing serious infective complications with current therapies.

MATERIALS AND METHODS

Data Collection

Clinical notes, flow charts, and computerized medical results from 280 patients with CLL who were followed at Birmingham Heartlands Hospital from 1995 to 2005 were reviewed retrospectively. The patient population included all consecutive patients who had a definitive diagnosis of CLL, as defined by standard morphologic and immunophenotypic criteria.15

Indications for treatment were those recommended by the National Cancer Institute Working Group (NCI-WG).16 Patients who required treatment were entered in the prevailing Medical Research Council trial whenever possible and, hence, received different initial therapies. In the CLL III trial, patients were assigned randomly to receive chlorambucil with or without epirubicin. In the CLL IV trial, patients were assigned randomly to receive chlorambucil versus fludarabine with or without cyclophosphamide. In the CLL pilot study, patients received fludarabine to maximal response followed by autologous HCT. Salvage therapies varied according to initial therapy and age but typically included fludarabine (with or without cyclophosphamide); cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP); alemtuzumab; and allogeneic HCT. Fludarabine refractoriness was defined as failure to respond to at least 2 cycles of a fludarabine-containing regimen or disease recurrence/progression within 6 months after the completion of chemotherapy.16Pneumocystis prophylaxis with cotrimoxazole was given routinely to patients receiving fludarabine or alemtuzumab and to patients who underwent allogeneic HCT. Cytomegalovirus antigenemia was monitored in patients who received alemtuzumab and in allogeneic HCT recipients. Selected patients with recurrent chest infections and concomitant lung conditions received prophylactic antibiotics.

Major infections were identified as those that required hospital admission and intravenous antibiotics. Infections that were treated with oral antibiotics were not considered in this study. Each infective episode was numbered chronologically, and a data series was collected. The causative organism, when identified, was classified as a bacterial, fungal, viral, parasitic, or clinical infection with negative culture. The infection site was classified as lower respiratory tract, upper respiratory tract, urinary tract, gastrointestinal tract, skin/soft tissue, central nervous system, and blood (septicemia). Fever of unknown origin (FUO) was defined as the presence of fever without an identifiable focus of infection and negative blood cultures.

Baseline Ig and creatinine levels were obtained. The glomerular filtration rate (GFR) was estimated by using the Modification of Diet in Renal Disease equation (GFR [mL/minute/1.73 m2] = 186.3 * SCr [mg/dL]− 1.154 * age [years]− 0.203 * 1.212 if black, * 0.742 if female).17 At the time of each infection, white blood cell and absolute neutrophil counts were recorded. Neutropenia was defined as an absolute neutrophil count <1.0 × 109/L.

Prognostic Factors

Peripheral blood samples were obtained before any therapy was instituted and after fully informed consent. Clinical staging was based on the Binet system.18 The mutational status of the IgVH gene was determined as described previously.19 Briefly, RNA was extracted by using the Qiagen RNeasy mini kit (Qiagen), and complementary DNA was synthesized by using the Reverse-iT kit (ABgene). The VH gene was amplified by using primers from the consensus sequences of the Framework 1 and joining region. The resulting polymerase chain reaction (PCR) products were sequenced and analyzed by using Immunoglobulin BLAST (National Center for Biotechnology Information, Bethesda, MD). Sequences with a germline homology ≥98% were considered unmutated, and those with a germline homology <98% were considered mutated.

Genetic analysis for the ataxia-telangiectasia mutated (ATM) and p53 genes was performed as reported previously.19, 20 Mutation analysis of the ATM coding sequence was performed by reverse transcriptase-PCR and restriction endonuclease fingerprinting or sequencing. Mutation analysis of the TP53 gene was performed by using amplification and sequencing of either the whole TP53 coding sequence or, alternatively, exons 3 through 10. A selected number of patients had cytogenetic analysis performed using fluorescent in situ hybridization (FISH), which was used to detect trisomy 12 and 6q and 11q, 13q, and 17p deletions (Vysis, London, United Kingdom), and/or conventional karyotype analysis.

To assess ZAP-70 expression by flow cytometry, peripheral blood cells were incubated with anti-CD3, CD5, CD19, and ZAP-70 (clone 1E7.2; Caltag) monoclonal antibodies. The cells were analyzed on a FACScalibur flow cytometer (Becton Dickinson), and at least 10,000 events were acquired from each sample. The percentage of ZAP-70-positive CLL cells in each sample was calculated by using the CD3-positive T-cell population to establish the lower limit of ZAP-70 expression. CD5-positive/CD19-positive cells above the CD3-positive threshold were considered ZAP-70-positive. The cut-off value for ZAP-70 positivity was 20%.4 To evaluate CD38 expression, triple-color immunophenotyping was performed using CD5, CD38, and CD19 monoclonal antibodies. The cut-off value for CD38 positivity was 30%.3, 21

Statistical Analysis

Baseline characteristics and prognostic factors were compared by using chi-square or Fisher exact tests for discrete variables and the Mann–Whitney test for continuous variables. The time to first infection (TTFI) and IRM were estimated as cumulative incidences considering noninfection-related deaths as a competing risk. Comparison of cumulative-incidence curves was performed by using Lunn–McNeil's approach, in which Cox regression is applied to competing risks. Several factors were analyzed for their association with the incidence of major infections, TTFI, and IRM, including age, gender, clinical stage, IgVH mutation status, ZAP-70 expression, CD38 expression, genetic abnormalities, Ig levels, and GFR. The effect of other variables, such as the number of therapy lines or response to therapy, was not evaluated, because they could not be known at diagnosis.22 The effect of the initial chemotherapy was analyzed in those patients who received treatment. For this particular analysis, the clock was set at the time of first treatment, not at diagnosis, and patients who developed any major infection before therapy was instituted were excluded from this separate analysis. In all statistical calculations, unadjusted P values <.05 were considered significant.

RESULTS

Baseline Characteristics

The characteristics of the study population are shown in Table 1. The median age was 69 years (range, 24–92 years). At presentation, 204 patients (73%) had Stage A CLL, and 76 patients (27%) had Stage B or C CLL. Twenty-seven patients (10%) developed immune cytopenias throughout the follow-up period, and 45 patients (16%) received corticosteroids or immunosuppressive agents for immune cytopenias or other medical conditions. Chemotherapy was required in 125 patients (45%) at a median of 4 months after diagnosis (range, 0–201 months). The median number of chemotherapy lines was 2 (range, 1–9 chemotherapy lines). Fifty-six patients (46%) received fludarabine as initial or salvage therapy, and 20 of those patients (36%) were refractory according to NCI-WG criteria. Sixteen patients (13%) underwent autologous HCT, and 5 patients (4%) underwent allogeneic HCT. The median follow-up for the whole population was 62 months (range, 0.4–302 months).

Table 1. Baseline Characteristics
CharacteristicNo. of patients (%)
  1. ATM indicates ataxia-telangiectasia mutated gene; IgVH: immunoglobulin VH;±, with or without; CHOP, cyclophosphamide, doxorubicin, vincristine, and prednisone.

Median age (range), y69 (24–92)
Gender: Male/female (%)58/42
Binet stage (n = 280 patients)
 Stage A204 (73)
 Stage B or C76 (27)
Lymphocyte doubling time (n = 266 patients) 
 >12 mo180 (68)
 <12 mo86 (32)
CD38 expression (n = 194 patients)
 Negative123 (63)
 Positive71 (37)
ZAP-70 expression (n = 151 patients)
 Negative66 (44)
 Positive85 (56)
Genetic abnormalities (n = 180 patients)
 Normal or 13q-125 (69)
 ATM24 (13)
 p537 (4)
 Other (+12, 6q-)24 (13)
IgVH mutation status (n = 231 patients)
 Unmutated52 (22)
 Mutated179 (78)
Treatment
 Chlorambucil ± prednisone93 (76)
 Fludarabine ± cyclophosphamide56 (46)
 CHOP18 (15)
 Alemtuzumab12 (10)
 Rituximab1 (1)
Median follow-up (range), mo62 (0.4–302)

The most common immune defect in our series was a low IgM level (<0.5 g/L), which was observed in 112 of 182 patients (62%). Low IgA (<0.8 g/L) and IgG (<6 g/L) levels also were detected in 66 patients (36%) and 53 patients (29%), respectively. The median calculated GFR was 63 mL per minute (range, 11–144 mL per minute). Nine patients (3%) with severe hypo-γ-globulinemia and recurrent infections received monthly Ig replacement (250 mg/kg), whereas 15 patients (6%) were on rotating prophylactic antibiotics, and 47 patients (17%) received cotrimoxazole thrice weekly for Pneumocystis prophylaxis.

Infective Episodes

After 1699 patient-years of observation, the incidence of major infections was 0.13 episodes per patient-year, which were distributed evenly across the entire period of follow-up. In total, 100 patients (36%) had at least 1 major infection that required hospital admission and intravenous antibiotics (median, 2 major infections; range, 1–8 major infections) over a median follow-up of 67 months. The respiratory tract was involved in 121 of 214 infective episodes (57%) (Table 2). Respiratory tract infections were less common in neutropenic patients (<1.0 × 109/L) (P = .002), whereas septicemias and episodes of FUO were more frequent in the presence of neutropenia (P<.001) and also were caused more frequently by Gram-positive bacteria compared with other infection sites (P = .018). There was a very significant association between infection sites and other comorbidities. Indeed, respiratory tract infections represented 85% of all episodes documented in patients with concomitant lung conditions, including chronic obstructive pulmonary disease, asthma, interstitial disorders, and lung carcinoma; whereas they constituted only 49% of all infections observed in patients without known lung conditions (P < .001). Conversely, patients with indwelling venous access were more likely to develop septicemia or FUO compared with patients who did not have such devices (P = .004).

Table 2. Characteristics of 216 Major Infections
InfectionNo of patients (%)
Causative organism
 Gram-positive bacteria56 (26)
  Streptococcus pneumoniae20 (9)
  Other streptococci6 (3)
  Enterococci6 (3)
  Staphylococci21 (10)
  Other gram-positive organisms3 (1)
 Gram-negative bacteria51 (24)
  Enteric rods24 (11)
  Pseudomonas spp.18 (8)
  Pleomorphic (Haemophilus spp., etc.)9 (4)
  Mycobacteria1 (1)
 Fungi11 (5)
  Yeasts (Candida spp., etc.)8 (3)
  Moulds2 (1)
  Pneumocystis jiroveci1 (1)
 Viruses16 (7)
  Herpesvirus13 (5)
  Respiratory virus2 (1)
  Polyomavirus1 (1)
 Unknown81 (37)
Site of infection
 Lower respiratory tract89 (41)
 Lower respiratory tract and septicemia29 (13)
 Upper respiratory tract5 (2)
 Urinary tract13 (6)
 Skin/soft tissues15 (7)
 Gastrointestinal tract4 (2)
 Central nervous system2 (1)
 Septicemia47 (22)
 Fever of unknown origin12 (6)
Neutrophil count
 >1.0 × 109/L147 (68)
 <1.0 × 109/L69 (32)

Appropriate cultures were obtained routinely in most major infections, and the causative organism was identified in 134 major infections (63%) (Table 2). Eighty-eight percent of those documented infections were bacterial, 7% were viral, and 5% were fungal. There was a clear association between some specific organisms and certain infection sites. For instance, Streptococcus pneumoniae and Haemophilus influenzae mostly were isolated in patients who had respiratory tract infections (80% and 89%, respectively). In contrast, other streptococci (excluding S. pneumoniae), enterococci, and staphylococci usually were identified in patients with septicemia (83%, 83%, and 61%, respectively). It is noteworthy that Pseudomonas spp. and enteric Gram-negative rods had no predilection for any particular infection site. Fungal infections were uncommon and were not associated significantly with neutropenia or a particular infection site. Most fungal infections were caused by yeasts (73%), although 2 infections were caused by moulds, and 1 infection was caused by Pneumocystis jiroveci (formerly known as Pneumocystis carinii). Viral infections usually were caused by herpesvirus (herpes simplex, varicella-zoster, and cytomegalovirus), although mild infections that did not require admission or intravenous therapy were not considered in the current study.

Several factors were associated with the number of infections per year. Patients with low IgG levels (P = .006), with low IgA levels (P = .008), and low IgM levels (P = .002) had an increased risk of infections. Other prognostic factors with influence on the infective risk were clinical Stage B/C disease (P < .001), positive CD38 expression (P < .001), abnormal genetics (P < .001), and mutated IgVH (P < .001). Other factors, such as renal function, age, and ZAP-70 expression, showed no association with the incidence of infection.

Time to First Infection

The 5-year risk for major infection was 29% (95% confidence interval [95% CI], 24–36%) for the whole population, and the median time to first major infection was 41 months (range, 0–191 months). The time to first major infection (TTFI) according to clinical stage, IgVH mutation status, CD38 expression, and genetic abnormality is shown graphically in Figure 1. Patients who were diagnosed in Stage B or C had a much shorter TTFI (median, 37 months) compared with patients who were diagnosed in Stage A (median, 52 months; P < .001). In addition, IgVH mutation status had a very significant effect on TTFI. Indeed, patients with unmutated genes had a median TTFI of 31 months, whereas patients with mutated genes doubled that figure (62 months; P < .001). In addition, patients with genetic abnormalities (p53, ATM, trisomy 12) also had a shorter TTFI (P < .001) along with patients who had CD38-positive cells (P < .001). It is noteworthy that other, “classic” predisposing factors, such as Ig levels and creatinine clearance, had no significant effect on the TTFI (Table 3). Finally, patients older than 70 years had a significantly shorter TTFI (P = .007) despite their much shorter follow-up (P < .001; Mann–Whitney test). Multivariate analysis showed that only clinical stage (P < .001) and mutation status (P = .010) had an independent effect on TTFI.

Figure 1.

The time to first major infection is illustrated according to (A) clinical stage, (B) immunoglobulin VH (IgVH) mutation status, (C) CD38 expression, (D) genetic abnormalities, and (E) initial therapy. In all analyses, noninfection-related deaths were considered as a competing risk. The effect of initial therapy was analyzed only in the group of patients who received treatment; it was measured from the time of first treatment (not from the time of diagnosis), and patients who developed a major infection before any therapy was instituted were excluded. ATM indicates ataxia-telangiectasia mutated gene.

Table 3. The Effect of Prognostic and Infection-Predisposing Factors on the Time to First Infection and Infection-Related Mortality
FactorTime to first infectionInfection-related mortality
HR (95% CI)Univariate PMultivariate PHR (95% CI)Univariate PMultivariate P
  • HR indicates hazards ratio; 95% CI, 95% confidence interval; NS, nonsignificant; Ig, immunoglobulin; CrCl, creatinine clearance.

  • *

    p53 Gene abnormalities, ataxia-telangiectasia mutated (ATM) gene abnormalities, or trisomy 12.

Gender NS  NS 
Age > 70 y1.75 (1.17–2.64).007NS3.13 (1.70–5.79)<.001<.001
Clinical Stage B or C4.14 (2.70–6.34)<.001<.0015.60 (3.00–10.47)<.001.001
CD38 positive3.30 (1.98–5.51)<.001NS4.05 (1.86–8.85)<.001NS
ZAP–70 positive NS  NS 
Abnormal genetics*3.16 (1.93–5.17)<.001NS2.99 (1.47–6.09).003NS
Unmutated IgVH gene4.37 (2.78–6.88)<.0010.013.30 (1.75–6.23)<.001NS
IgG <5 g/L NS  NS 
IgA <0.8 g/L NS  NS 
IgM <0.5 g/L2.0 (1.16–3.45).013NS NS 
CrCl <60 mL/min NS  NS 
First treatment other than chlorambucil2.73 (1.60–4.67)<.001  NS 

A separate analysis was performed for the patients who required therapy. It is worth noting that the drug chosen as first-line therapy had a significant impact on TTFI. Indeed, patients who received chlorambucil were less prone to major infectious episodes compared with patients who received fludarabine with or without cyclophosphamide or CHOP (P < .001) (Fig. 1E). It also is interesting to note that both groups (chlorambucil vs. other) were balanced well in terms of prognostic factors (IgVH mutation status, Binet stage, CD38 status, genetic abnormalities), Ig levels, and renal function (Table 4).

Table 4. Prognostic and Infection-Predisposing Factors According to Initial Therapy*
CharacteristicNo. of patients (%)P
ChlorambucilFludarabine ± Cyclophosphamide or CHOP
  • ± Indicates with or without; CHOP, cyclophosphamide, doxorubicin, vincristine, and prednisone; NS, nonsignificant; Ig, immunoglobulin.

  • *

    Patients who never received any therapy or who had a major infection before initial therapy was started were excluded from this analysis.

  • P values were determined by using the chi-square test or the Fisher exact test.

  • p53 Gene abnormalities, ataxia-telangiectasia mutated (ATM) gene abnormalities, or trisomy 12.

Age >70 y46 (56)4 (13)<.001
Clinical Stage B or C47 (57)18 (58)1.0 (NS)
CD38 positive35 (55)11 (61).789 (NS)
ZAP-70 positive40 (63)13 (72).580 (NS)
Genetic abnormalities29 (45)12 (52).629 (NS)
Unmutated IgVH gene30 (40)11 (42).820 (NS)
IgG <6 g/L26 (43)9 (33).483 (NS)
IgA <0.8 g/L37 (61)15 (56).814 (NS)
IgM <0.5 g/L43 (71)20 (74).802 (NS)

Recurrent Infective Episodes and Infection-Related Mortality

Seventy-eight patients (78%) survived their first major infective episode. The only factor that was associated significantly with the occurrence of further infections was clinical staging, in that 89% of patients who were diagnosed in Stage B/C had at least a second major infection compared with 51% of patients who were diagnosed in Stage A (P < .001). No other factor, including CD38 expression, IgVH status, genetic abnormalities, or Ig levels, had a significant effect on the risk of secondary infections.

With a median follow-up of 5 years (range, from 1 month to 25 years), 86 patients (31%) died. Infections were the most common cause of death, accounting for 51% of all fatalities. Other relatively common causes of death were cardiovascular complications (9%), other malignancies (8%), and CLL progression/Richter transformation (6%). The 5-year IRM rate was 8% (95% CI, 5–13%).

Figure 2 shows IRM graphically according to clinical stage (P < .001), IgVH mutation status (P < .001), CD38 expression (P < .001), and genetic abnormalities (P = .003). As expected, the effect of age was also quite strong (P < .001). Multivariate analysis confirmed the independent effect of age (P < .001) and stage (P = .001) on IRM. It is noteworthy that first-line chemotherapy did not have any significant impact on IRM when treated patients were analyzed separately (Table 3).

Figure 2.

Infection-related mortality is illustrated according to (A) clinical stage, (B) immunoglobulin VH (IgVH) mutation status, (C) CD38 expression, and (D) genetic abnormalities. In all analyses, noninfection-related deaths were considered as a competing risk. ATM indicates ataxia-telangiectasia mutated gene.

DISCUSSION

Infective complications are having a greater impact on the outcome of patients with CLL as treatment alternatives are becoming increasingly complex and aggressive.23–25 The current study provides a comprehensive analysis of major infective complications in a large group of unselected patients with CLL. Appropriate cultures were obtained in every major infective episode, which may explain the high percentage of isolated organisms (63%) compared with other series.7, 10, 23, 25

Several predisposing factors have been suggested over the last 50 years, including hypo-γ-globulinemia, neutropenia, defective T-cell function, and reduced complement activity.7, 26 In keeping with previous reports, the infective pattern observed in the current work was very similar to that observed in primary humoral immune deficiencies.7, 27, 28 Encapsulated organisms (Streptococcus pneumoniae, Haemophilus influenzae) were common pathogens in respiratory tract infections, whereas other streptococci and staphylococci were identified more frequently in episodes of septicemia. Other frequent isolates were Gram-negative bacteria, particularly enteric rods and Pseudomonas spp. We did not observe an excessive incidence of fungal or parasitic infections, possibly because only a small proportion of patients received intensive therapies. Major viral infections also were relatively uncommon and mostly were caused by herpesvirus, as described previously.10, 23

The median age at diagnosis in the current study was 69 years, and elderly patients often have multiple medical conditions. It is difficult to analyze the effect of other medical conditions on the infective risk of patients with CLL, but we have observed a significantly distinct infective pattern in 2 important patient groups. Thus, patients with end-stage renal disease and permanent indwelling catheters had significantly more episodes of septicemia compared with patients who did not have such devices. In addition, patients with chronic lung diseases had more frequent chest infections than patients without any known lung condition. It could be argued that chronic lung disease, particularly bronchiectasis, is a common medical problem in patients with primary humoral immune deficiencies, leading to recurrent chest infections and even severe respiratory impairment.27 Indeed, there were certain patients in whom it was difficult to determine whether the chronic lung disorder was a preexisting condition or was caused by CLL-associated hypo-γ-globulinemia. However, we believe that concomitant medical conditions should be taken into account to predict the infective risk and to consider appropriate Ig or antibiotic prophylaxis.

Our results indicated that several prognostic factors, particularly clinical stage, CD38 expression, genetic analysis, and IgVH mutation status, all were powerful predictors for the incidence of infection, TTFI, and IRM (see Figs. 1, 2). Clinical staging using either the Rai system or the Binet system has allowed patients' stratification for therapy and clinical research for over 30 years.18 However, it cannot distinguish rapidly progressing patients from patients whose disease remains quiescent for a long period.2 The prognostic significance of IgVH mutation status has been confirmed repeatedly by many independent groups3, 5, 29 and in our own work.30 Patients with CLL with malignant cells that have unmutated IgVH genes have a shorter overall and treatment-free survival and have a worse response to therapy. This mutational status is the only prognostic factor that is intrinsic to the malignant clone and can be determined regardless of whether the patient has received therapy.2 CD38 expression initially was suggested as a surrogate for IgVH mutation status.3 Subsequent studies did not confirm this observation but established its independent prognostic value.21, 29 Unfortunately, those studies also showed differences in CD38 expression for individual patients over time, limiting its value as a prognostic factor.21, 29 Genomic abnormalities, specifically 11q and 17p deletions that affect the ATM and TP53 genes, respectively, have been associated consistently with shorter survival.6, 29 However, CLL cells can acquire chromosomal aberrations during the disease course in a phenomenon known as clonal evolution.31 Conversely, abnormal p53 function is present in a larger proportion of patients with CLL (26%) than chromosome 17 abnormalities by FISH analysis (7%), and it is not clear which test provides the best prognostic information.32

In our study, IgVH mutation status had a very significant impact on overall and treatment-free survival (data not shown) and also on the incidence of infection, TTFI and IRM. It is noteworthy that patients who had unmutated IgVH genes did not have a higher incidence of hypo-γ-globulinemia or receive different initial therapy compared with patients who had mutated IgVH (data not shown); however, they experienced more frequent and lethal infections. The reason for this association remains unclear. There is growing evidence that a significant proportion of patients with CLL who have unmutated IgVH genes have almost identical antigen receptors, suggesting that their malignant cells are selected by specific antigens.33, 34 It is possible that patients with these “antigen-driven” diseases have a reduced ability to develop immune responses against different bacterial or viral antigens. Alternatively, the IgVH mutation status simply may reflect the degree of disease aggressiveness and immune deficiency. Indeed, a great proportion of patients with unmutated IgVH genes require chemotherapy, which also may increase their risk of infections.23, 25

It is surprising that ZAP-70 expression was not predictive of infection risk, given the reported strong correlation between ZAP-70 expression and IgVH mutation status.4 Our correlation was around 60%, clearly below most reported series,4 implying that our ZAP-70 results must be regarded with caution. We currently are reanalyzing our series with a different monoclonal antibody.

We also assessed the effect of first-line therapy on TTFI and IRM. Our study confirmed that patients who received fludarabine-containing regimens or CHOP as initial therapy had a shorter TTFI than patients who received chlorambucil,25 although they were significantly younger (Table 4). However, this increased risk of major infections did not translate into a higher IRM rate and, thus, had little impact on overall survival. These results should be interpreted with caution, because this was a retrospective analysis, even though many of these patients were recruited into several randomized clinical trials. Fortunately, both therapy groups were balanced well in terms of all prognostic factors available except age (Table 4), which confirms the previous observation that initial therapy affects the infective risk.23 We did not analyze the effects of the number of chemotherapy regimens or response to therapy, because they were conditioned by the aggressiveness of the disease, and our time-dependent models did not allow for the inclusion of variables that were unknown at the time of diagnosis.25 However, it must be emphasized that response to therapy seems to be associated with the infective risk of patients with CLL who are treated with fludarabine.35, 36

In CLL, hypo-γ-globulinemia is a continuous process that occurs spontaneously throughout the natural course of the disease and is related to disease duration and stage.7, 9 Low IgG and IgA levels have been associated with increased frequency and severity of infections and shorter overall survival.9, 28 We considered only Ig determinations that were made at diagnosis to avoid any bias caused by disease duration. Ig levels were associated inversely with the incidence of infections, as expected, but IgG and IgA levels had no impact on TTFI, IRM, or even overall and treatment-free survival (data not shown). IgM levels had a marginal impact on TTFI in the univariate analysis, but they were reduced significantly in patients with unmutated IgVH genes; thus, and their true predictive value is difficult to assess. These results are in keeping with recent studies and suggest that Ig levels are becoming less important as more aggressive therapies are used.23, 25 Furthermore, Ig levels measured at diagnosis were unable to predict patient survival and should be monitored regularly, because they change throughout the disease course.

Equally, baseline renal function had no effect on TTFI or IRM. This contrasts with a study published by Anaissie et al.,10 who reported that a serum creatinine level >1.4 mg/dL (>124 μM) was a significant risk factor for major infections in a group of patients with CLL who were treated with fludarabine. It is noteworthy that this phenomenon was apparent only in those patients who received fludarabine plus chlorambucil in a different study.23 We included both treated and untreated patients, and patients who had a GFR <30 mL per minute were not eligible for any fludarabine-containing regimen, which may explain the lack of effect of renal function on TTFI or IRM. Conversely, renal function tends to change over the course of the disease as much as Ig levels, limiting its use as a predictive factor.

There are several implications of these results in terms of prophylaxis. Ig replacement therapy is recommended after the first or second major infective episode in patients with hypo-γ-globulinemia.26 This approach stems from a randomized trial in which regular Ig infusions reduced the incidence of bacterial infections in patients who had hypo-γ-globulinemia and a history of infections.37 We believe that Ig replacement therapy still is appropriate in these patients,26 because they had an increased incidence of major infections, and this approach may reduce the number of hospital admissions. Furthermore, the infective pattern observed (frequent encapsulated organisms, fewer viral and fungal infections) also supports the use of regular Ig infusions. However, we also believe that a few other factors should be taken into account, particularly disease stage and IgVH mutation status. Patients who were diagnosed in Stage B or C or who had unmutated IgVH genes show a significantly increased risk of recurrent infections and should be considered for prophylaxis shortly after the first episode. We also recommend Pneumocystis prophylaxis in patients who receive fludarabine, alemtuzumab, or HCT and antibiotic prophylaxis in selected patients with recurrent infections and chronic lung conditions, all in keeping with recently published guidelines.26

From the current results, we conclude that clinical stage at diagnosis, IgVH mutation status, and initial therapy are possible predictors of severe infections in patients with CLL. IgVH mutation status is particularly useful, because it is the only factor that is intrinsic to the malignant clone, it does not change throughout the course of the disease, and it is very reproducible. Other prognostic markers, such as CD38 expression and genetic abnormalities, also had a significant impact on TTFI and IRM, but they may change over time, and optimal cut-off levels or the appropriate type of test (FISH vs. PCR) have not been established firmly. This was a retrospective analysis, and its conclusions should be regarded with caution; however, we hope it will help to identify CLL patients who are at particularly high risk of serious infections and should be considered for immunoglobulin or antibiotic prophylaxis.

Acknowledgements

We thank Drs. Belinda Austen and Tatjana Stankovic for performing the ATM and TP53 gene analysis.

Ancillary