Hypogammaglobulinemia in Newly Diagnosed Chronic Lymphocytic Leukemia: Natural History, Clinical Correlates and Outcomes

Abstract

Background

Although hypogammaglobulinemia is a well-recognized complication in chronic lymphocytic leukemia (CLL), its prevalence at the time of CLL diagnosis, and association with novel prognostic markers and clinical outcome is not well understood.

Methods

All patients seen at Mayo Clinic between 1/1999–7/2013 with newly diagnosed CLL and who had baseline assessment of serum immunoglobulin G (IgG) were included. The relationship between hypogammaglobulinemia at diagnosis and novel prognostic parameters, time to first treatment (TFT) and overall survival (OS) were evaluated.

Results

Of 1485 patients who met the eligibility criteria, 382 (26%) had hypogammaglobulinemia (median IgG=624 mg/dL) while the remaining 1103 (74%) had normal serum IgG levels (median IgG=1040 mg/dL). Patients with hypogammaglobulinemia at diagnosis were more likely to have advanced Rai stage (III–IV, p=0.001) and have higher expression of CD49d (p<0.001) compared to patients with normal IgG. Although the median TFT for patients with hypogammaglobulinemia was shorter compared to patients with a normal IgG (3.8 years vs. 7.4 years, p<0.001) on multivariable analysis, there was no difference in OS between these two groups (12.8 years vs. 11.3 years, p=0.73). Of 1103 CLL patients who had a normal IgG at diagnosis and who did not receive CLL therapy, the risk of acquired hypogammaglobulinemia was 11% and 23% at 5 and 10 years, respectively.

Conclusion

Hypogammaglobulinemia is present in one-quarter of newly diagnosed CLL patients. Approximately one-quarter of CLL patients with normal IgG at diagnosis will subsequently develop hypogammaglobulinemia on long-term follow-up. The presence of hypogammaglobulinemia does not appear to impact survival.

INTRODUCTION

Since the original description of hypogammaglobulinemia in a CLL patient in 1956,¹ our understanding of the spectrum of immune dysregulation in CLL has evolved. Although hypogammaglobulinemia is the most commonly described immune defect in CLL, abnormalities in T-cells,^2–4 the complement system,^{5, 6} and neutrophil and monocyte function^{7, 8} also contribute to infectious complications. With the introduction of highly effective chemotherapy^9–11 and chemoimmunotherapy^{12, 13} regimens for the treatment of CLL, immune deficiency as a consequence of therapy has also increased the risk of infections.

Hypogammaglobulinemia is thought to occur due to defective functioning of the non-clonal CD5-negative B-cells,¹⁴ and is more pronounced with prolonged disease duration and advanced disease stage.^{15, 16} Although prior studies have reported the frequency of hypogammaglobulinemia in CLL patients to range from 20–70%,^16–23 they are limited by the heterogeneity of the patient population studied (previously untreated and treated CLL patients) and lack of sufficient follow-up to determine the risk of future hypogammaglobulinemia in patients with normal immunoglobulin levels at diagnosis. The relationship between hypogammaglobulinemia, novel prognostic markers (such as immunoglobulin heavy chain gene mutation status [IGHV], genetic abnormalities detected by fluorescence in situ hybridization [FISH], expression of CD38, CD49d and zeta-associated protein 70 [ZAP-70], stereotyped B-cell receptors), and time to first treatment (TFT) have also not been explored systematically. There are conflicting reports on the impact of hypogammaglobulinemia on overall survival (OS), with some studies suggesting that it is associated with worse survival^{15, 16} and others not.^{17, 24, 25}

We used the Mayo Clinic CLL database to assess the prevalence of hypogammaglobulinemia in a large cohort of newly diagnosed CLL patients. Additionally, we assessed the impact of hypogammaglobulinemia on TFT and OS in all patients. Finally, we determined the incidence of hypogammaglobulinemia over time in newly diagnosed CLL patients who had normal immunoglobulin levels at diagnosis.

METHODS

Patients

The Mayo Clinic CLL database includes all patients with a pathologic diagnosis of CLL that were seen in the Division of Hematology, Rochester, MN and who permit their records to be used for research purposes.^26–28 Baseline factors abstracted from clinical records on all patients include age, sex, Rai stage, complete blood count (CBC), absolute lymphocyte count (ALC), IGHV mutation status, cytogenetic abnormalities detected by interphase FISH, and expression of ZAP-70, CD49d and CD38. The treatments received, complications of therapy, TFT and OS are also recorded for all patients.

CLL patients seen at Mayo Clinic from January 1999 through July 2013 were eligible to be included in this study if: 1) they were seen at Mayo Clinic within 12 months of their CLL diagnosis, 2) they had not received treatment for CLL at time of initial evaluation, and 3) serum immunoglobulin levels were analyzed within 90 days of their initial visit at Mayo Clinic. For patients in whom more than one serum IgG level was available within 90 days, the value closest to the date of their initial visit was used for the purposes of this study. Serum immunoglobulins were quantitated by radial immunodiffusion using Immunoplates (Behring Institute, Marburg, FRG) and/or serum protein electrophoresis.

In our laboratory, the normal value for serum IgG is 757–1590 mg/dL. CLL patients with serum IgG <757 mg/dL were classified as having hypogammaglobulinemia. For the present analysis, patients with IgG<757 mg/dL were further stratified by terciles. The relationship between hypogammaglobulinemia at diagnosis and prognostic parameters including IGHV mutation status, genetic abnormalities detected by FISH, and ZAP-70, CD38 and CD49d expression was assessed. The association of hypogammaglobulinemia with TFT and OS was also evaluated for all patients. The time to development of hypogammaglobulinemia over long-term follow-up was also determined for patients with a normal IgG level at CLL diagnosis who had at least one additional IgG level measured at least six months after the initial level. The clinical and biologic factors associated with a higher risk of subsequent hypogammaglobulinemia in these patients were analyzed.

Statistical Analysis

We used Chi-square and Fisher’s exact tests to compare discrete and ordinal variables, respectively, and the Kruskal Wallis test to compare continuous variables. TFT was defined as the interval between the diagnosis of CLL and the initiation of first treatment for CLL. OS was defined as the interval between the diagnosis of CLL and death due to any cause. Time to development of hypogammaglobulinemia was defined as the interval between CLL diagnosis and the first identification of low IgG level during follow-up. Death, loss to follow-up and receipt of CLL therapy were considered as censoring variables. Kaplan-Meier plots were generated to depict time to hypogammaglobulinemia, and display TFT and OS by hypogammaglobulinemia categories. Multivariable Cox proportional hazards regression analysis was used to determine factors which predicted TFT and OS in the entire cohort. Among patients with a normal IgG at CLL diagnosis, multivariable Cox proportional hazards regression analysis was used to determine factors which predicted time to hypogammaglobulinemia. Due to missing data, multiple models were run; introducing one factor at a time (such as IGHV, FISH, CD38, CD49d, ZAP-70, and serum IgG) to the base model (which consisted of age, sex and Rai stage). All statistical analyses were performed using SAS 9.3 (SAS Institute, Cary, NC).

RESULTS

Hypogammaglobulinemia at Diagnosis

Two thousand and thirty CLL patients were seen at Mayo Clinic, Rochester, MN during the study period. Of these, 1485 CLL patients who were both previously untreated and had serum IgG level within 90 days of their initial visit were included in this study. Of these patients, 382 (26%) had hypogammaglobulinemia at the time of CLL diagnosis (median serum IgG=624 mg/dL; range: 111–755 mg/dL), while the remaining 1103 (74%) patients had normal serum IgG levels (median serum IgG=1040 mg/dL; range 757–4000 mg/dL). Baseline demographic and clinical characteristics of patients with and without hypogammaglobulinemia are shown in Table 1. Patients with hypogammaglobulinemia were more likely to have advanced Rai stage (III–IV, p=0.001) and have higher expression of CD49d (p<0.001) compared to patients with normal IgG levels at CLL diagnosis.

Table 1.

Baseline characteristics of patients with and without hypogammaglobulinemia at CLL diagnosis

Characteristic		Hypogammaglobulinemia at diagnosis, (%) or [range]	Normal IgG at diagnosis (%) or [range]	p-value
Number of patients		382	1103
Median serum IgG level (mg/dL)		624 [111–755]	1040 [757–4000]
Median age, years		62 [37–94]	64 [27–94]	0.11
Male		268 (70)	716 (65)	0.06
Rai Stage	Low risk (0)	187 (49)	620 (56)	0.001
	Intermediate risk (I–II)	160 (42)	433 (39)
	High risk (III–IV)	32 (8)	45 (4)
IGHV mutation*	Mutated	140 (54)	391 (58)	0.21
	Unmutated	121 (46)	281 (42)
ZAP-70*	Negative (<20%)	185 (61)	505 (67)	0.08
	Positive (≥20%)	119 (39)	254 (33)
CD38*	Negative (<30%)	246 (70)	703 (72)	0.37
	Positive (≥30%)	106 (30)	268 (28)
CD49d*	Negative (<30%)	135 (57)	440 (75)	<0.0001
	Positive (≥30%)	103 (43)	150 (25)
FISH*	deletion13q	124 (42)	328 (40)	0.32
	normal	66 (22)	232 (28)
	+12	61 (21)	140 (17)
	deletion11q	30 (10)	69 (8)
	deletion17p	11 (4)	39 (5)

^*Not available for all patients

Table 2 shows the baseline characteristics of patients with hypogammaglobulinemia (serum IgG <757 mg/dL, n=382) when they were sub-stratified into terciles by immunoglobulin level. Patients in the lower tercile (serum IgG <585 mg/dL, n=127) were more likely to have advanced Rai stage (III–IV) at diagnosis (15% vs. 6% vs. 4%, p=0.008), be CD49d positive (56% vs. 43% vs. 32%, p=0.01), and have trisomy 12 on FISH testing (29% vs. 23% vs. 9%, p=0.005) compared to patients in the middle (serum IgG:585–687 mg/dL, n=127) and upper (serum IgG:688–756, n=128) terciles.

Table 2.

Baseline characteristics of CLL patients with hypogammaglobulinemia that is subdivided by three terciles

Characteristic		Lower Tercile IgG <585 mg/dL (%) or [range]	Middle Tercile IgG: 585–687 mg/dL (%) or [range]	Upper Tercile IgG: 688–756 mg/dL (%) or [range]	p- value
Number of patients		127	127	128
Median serum IgG, mg/dL		500 [111–584]	623 [585–687]	711 [688–755]
Median age, years		62 [39–90]	62 [41–92]	62 [37–94]	0.8
Male		96 (76)	87 (68)	85 (66)	0.24
Rai Stage	Low risk (0)	52 (42)	63 (50)	72 (57)	0.01
	Intermediate risk (I–II)	54 (43)	56 (44)	50 (39)
	High risk (III–IV)	19 (15)	8 (6)	5 (4)
IGHV mutation	Mutated	40 (47)	52 (58)	48 (56)	0.32
	Unmutated	45 (53)	38 (42)	38 (44)
ZAP-70	Negative (<20%)	54 (54)	59 (60)	72 (68)	0.13
	Positive (≥20%)	45 (45)	40 (40)	34 (32)
CD38	Negative (<30%)	75 (65)	79 (67)	92 (78)	0.06
	Positive (≥30%)	41 (35)	39 (33)	26 (22)
CD49d	Negative (<30%)	34 (44)	47 (57)	54 (68)	0.01
	Positive (≥30%)	43 (56)	35 (43)	25 (32)
FISH	deletion13q	35 (35)	46 (46)	43 (44)	0.06
	normal	19 (19)	19 (19)	28 (29)
	+12	29 (29)	23 (23)	9 (9)
	deletion11q	12 (12)	7 (7)	11 (11)
	deletion17p	4 (4)	2 (2)	5 (5)

Hypogammaglobulinemia at Diagnosis and Clinical Outcome

After median follow-up of 4.2 years, 479 (32%) patients had received treatment for CLL, and 328 (22%) patients had died. The median TFT for patients with hypogammaglobulinemia was 3.8 years compared to 7.4 years for patients with a normal IgG (p<0.001, Figure 1A). Among patients with hypogammaglobulinemia, the median TFT for patients in the lower, middle, and upper terciles were 1.7, 3.9, and 6.3 years respectively (p<0.001, Figure 1B). On multivariable analysis which included age, gender, Rai stage, IGHV mutation status, FISH risk category and hypogammaglobulinemia, the presence of hypogammaglobulinemia was independently associated with a shorter TFT (HR=1.5, 95%CI=1.1–2.2; p=0.02; Table 3).

Figure 1.

A: Time to first treatment in CLL patients with hypogammaglobulinemia at diagnosis compared to normal IgG at diagnosis

B: Time to first treatment among CLL patients with hypogammaglobulinemia according to their serum IgG level at diagnosis

Table 3.

Multivariable model to predict time to first CLL therapy and overall survival

Characteristic		TFT		OS
		HR (95% C.I.)	p-value	HR (95% C.I.)	p-value
Age1		1.0 (0.99–1.0)	0.56	1.1 (1.0–1.1)	<0.0001
Male Sex		1.6 (1.1–2.3)	0.01	1.1 (0.6–2.1)	0.66
Rai Risk Category
Intermediate (stage I/II)2		3.0 (2.1–4.4)	<0.0001	1.7 (0.9–3.1)	0.08
High (stage III/IV)2		15.3 (8.4–28.0)	<0.0001	6.1 (2.1–17.9)	0.001
IGHV unmutated		2.6 (1.8–3.8)	<0.0001	1.9 (0.99–3.7)	0.06
FISH risk category
deletion 11q/17p3		1.4 (0.9–2.3)	0.13	3.7 (1.7–7.8)	0.001
trisomy 123		1.0 (0.7–1.5)	0.96	1.8 (0.9–3.8)	0.12
Hypogammaglobulinemia		1.5 (1.1–2.2)	0.02	0.9 (0.5–1.6)	0.67

¹for each year older

²relative to Rai 0

³relative to no detectable FISH abnormality or del13q as the sole abnormality

Abbreviations Used:

TFT: Time to first therapy; OS; overall survival; HR=hazard ratio; C.I. =confidence interval

The median OS of patients with hypogammaglobulinemia was 12.8 years compared to 11.3 years in those with a normal IgG level (p=0.73, Figure 2A). The impact of low serum IgG levels on OS remain unchanged when the cutoff of serum IgG for a diagnosis of hypogammaglobulinemia was decreased from 767 mg/dL (lower limit of normal in our laboratory) to 500 mg/dL (data not shown). Among patients with hypogammaglobulinemia, there was no difference in OS among patients in the lowest tercile compared to those in the middle and upper terciles (10.1 years vs. not reached vs. 10.2 years, p=0.09, Figure 2B). On multivariable analysis (which included the variables included in the TFT model), hypogammaglobulinemia was not associated with OS (HR=0.9, 95% CI=0.5–1.6; p=0.67, Table 3).

Figure 2.

A: Overall survival of CLL patients with hypogammaglobulinemia compared to those with normal IgG levels at diagnosis

B: Overall survival of CLL patients with hypogammaglobulinemia according to their serum IgG level at diagnosis

Development of Hypogammaglobulinemia during Course of Disease

Of 1103 CLL patients who had a normal IgG at diagnosis, 405 (37%) patients had a subsequent IgG measured at least 6 months after diagnosis (clinically ascertained during routine care), and 698 (63%) patients did not have a repeat serum IgG assessment. There were no differences in the baseline characteristics (including age, sex, Rai stage, expression of CD38, CD49d, ZAP-70, IGHV mutation status and FISH category) between patients who had a repeat IgG assessment versus those who did not (data not shown). The median follow-up was longer for patients that had a repeat IgG measurement (6.3 years) compared those who did not have a repeat IgG measured (2.8 years; p<0.001). Clinically ascertained hypogammaglobulinemia among all 1103 patients with initially normal IgG at CLL diagnosis occurred in 18% of patients at 5 years and 32% at 10 years. Among untreated patients, the rate of clinically ascertained hypogammaglobulinemia at 5 and 10 years were 11% and 23% respectively (Figure 3). After adjusting for age, sex, and Rai stage, factors associated with risk of developing hypogammaglobulinemia during follow-up among patients with initially normal immunoglobulin levels were unmutated IGVH (HR=1.9, 95% CI=1.2–3.0; p=0.03) and high CD49d expression (HR=1.8, 95% CI=1.1–3.1; p=0.005).

Figure 3.

Risk of acquired hypogammaglobulinemia in patients with normal immunoglobulin levels at CLL diagnosis (n=1103). Solid line represents risk for all patients, and dashed line represents risk for patients who did not receive treatment for CLL.

DISCUSSION

This large retrospective cohort study comprehensively describes the prevalence of hypogammaglobulinemia and the association of serum IgG levels at the time of CLL diagnosis with novel prognostic parameters and clinical outcomes. The results of our study indicate that: 1) ~25% of newly diagnosed CLL patients have hypogammaglobulinemia at diagnosis; 2) ~25% patients with normal IgG levels at CLL diagnosis will subsequently develop hypogammaglobulinemia over the next 10 years, without receiving CLL-specific therapy; 3) the median time to first CLL therapy is shorter among patients with hypogammaglobulinemia, independent of other prognostic markers at diagnosis; and 4) hypogammaglobulinemia at diagnosis has no apparent relationship with OS.

How do these findings relate to previous studies? Several reports have described the prevalence of hypogammaglobulinemia to range from 20–70% among heterogeneous cohorts of previously untreated and treated CLL patients (collectively representing ~900 patients).^16–23 A recent publication from the Israeli CLL group of 857 Binet Stage A CLL patients (both previously treated and untreated) reported a low serum IgG level in ~11% of patients.²⁵ Rozman et al and Davey et al reported hypogammaglobulinemia in 10–44% of previously untreated CLL patients (total n for both studies=275).^{15, 29} To the best of our knowledge, the current study represents the largest cohort (n=1485) of CLL patients in whom serum immunoglobulin levels were available at diagnosis, and indicates that hypogammaglobulinemia is present in ~25% patients. Although consistent with prior reports, our results are in contrast to those reported by Tsai and colleagues who analyzed stored serum samples of 45 individuals who later developed CLL among 77,469 healthy adults enrolled in the US Prostate, Lung, Colorectal, and Ovarian (PLCO) screening trial. They found only 1/45 (3%) patient had hypogammaglobulinemia (detected 3 years prior to CLL diagnosis); thereby suggesting that hypogammaglobulinemia rarely predates onset of CLL.³⁰

In our study, there were no significant differences in immunoglobulin levels by age, sex, IGHV mutation status, expression of CD38 and ZAP-70, and FISH risk category at CLL diagnosis. However, patients with advanced Rai (III–IV) stage and positive CD49d expression were more likely to have hypogammaglobulinemia at diagnosis. CD49d, the α4 subunit of the α4b1 integrin heterodimer, plays an important role in leukocyte trafficking,³¹ facilitates interaction between leukocytes and stromal cells in the marrow via vascular cell adhesion molecule-1 (VCAM-1) and fibronectin,³¹ and also serves as a signaling receptor that influences B-cell survival via upregulation of Bcl-2 family members.³² A high expression of CD49d has been shown to independently predict poor survival in CLL patients.^{33, 34} The precise mechanism by which alteration in CD49d expression and/or trisomy 12 may influence serum immunoglobulin levels remains unclear at this time.

Our results also indicate the presence of low serum IgG level at the time of CLL diagnosis is an independent predictor of time to first CLL treatment. A dose effect was observed between serum IgG levels and TFT, with the shortest TFT among patients with the lowest immunoglobulin levels. These findings indicate that the presence of hypogammaglobulinemia at the time of CLL diagnosis may be a marker of a more biologically aggressive clone.³⁵ Given the retrospective nature of this clinical study, we are unable to perform correlative studies to comprehensively identify other immune function abnormalities (such as the number and function of T-cell, NK-cell, and normal B-cells), that may account for the observed short TTT in patients with hypogammaglobulinemia. Our suspicion is that the presence of hypogammaglobulinemia may reflect a more aggressive CLL clone, where hypogammaglobulinemia is a marker rather than mechanistically related to disease progression. Our findings are in contrast to a recent study reported by Shvidel et al,²⁵ where serum IgA levels (but not serum IgG) correlated with a shorter TTT in a group of 857 Binet Stage A CLL patients. Several studies have shown that male CLL patients have a worse OS compared to females.^36–38 Wierda and colleagues reported that among 930 previously untreated CLL patients, male sex was associated with a shorter TTT in univariable analysis.³⁹ The finding of shorter TTT among male CLL patients on multivariable analysis has not been reported before, and needs replication in other cohorts of patients. In contrast to its impact on TFT, hypogammaglobulinemia had no relationship to overall survival in our study.

Very few studies have described the incidence of hypogammaglobulinemia during long-term follow-up of newly diagnosed CLL patients with normal immunoglobulin levels at diagnosis. Rozman et al reported that of 247 newly diagnosed CLL patients, 48 (19%) had hypogammaglobulinemia at diagnosis. Of the remaining 199 patients who had normal immunoglobulin levels, approximately 40% developed hypogammaglobulinemia up to 9 years after CLL diagnosis and before receiving CLL therapy.¹⁵ The results from our current study suggest that ~25% of patients with initially normal immunoglobulin levels develop hypogammaglobulinemia 10 years after diagnosis, without receiving CLL-specific therapy. Features of high-risk CLL such as unmutated IGHV and positive expression of CD49d at CLL diagnosis were associated with a higher risk of future hypogammaglobulinemia.

Traditional chemotherapy and chemoimmunotherapy approaches for CLL tend to decrease rather than increase immunoglobulin levels and can magnify immunosuppression.^{12, 13} In this regard it should be emphasized that the iwCLL/NCI working group guidelines for patients with CLL explicitly state that hypogammaglobulinemia should not be considered an indication for treatment of CLL.^{40, 41} In contrast to traditional chemotherapy-based treatments, novel agents such as ibrutinib (a Bruton’s tyrosine kinase inhibitor) and idelalisib (a phosphatidylinositide 3-kinase δ inhibitor inhibitor), that have demonstrated excellent efficacy in the treatment of CLL do not seem to exacerbate the risk of hypogammaglobulinemia.^{42, 43} In fact, among 31 elderly patients with previously untreated CLL, ibrutinib monotherapy improved serum IgA and IgM levels, and had no significant impact on serum IgG levels after a median of 21 months (range, 14.8–22.6 months) of therapy.⁴⁴ Although data from the use of these novel therapies are promising, both with respect to their efficacy in the treatment of CLL and restoring the deranged immune system that accompanies CLL, longer follow-up will be required to accurately assess their impact on infectious complications.

There are several limitations in our study. Our report is a single institution study and the findings may not be generalizable. Since all the novel prognostic markers were not available during the entire study period, we do not have complete information for all patients. Due to the referral nature of practice at our tertiary care institution, we did not have reliable information about infectious complications and the use of intravenous immunoglobulin for the management of hypogammaglobulinemia in this cohort of patients. Given that only ~40% patients with normal IgG at diagnosis in our study had a subsequent clinically ascertained serum IgG, we are unable to accurately define the risk of acquired hypogammaglobulinemia. Also, among patients with a normal IgG at diagnosis, the median duration of follow-up was longer among those who had a subsequent IgG level measured compared to those who did not, suggesting that those with longer follow-up were more likely to have had repeat testing. Finally, there are conflicting reports on the importance of IgG subclass (IgG1, IgG2, IgG3, and IgG4) deficiency and the risk of infectious complications in CLL patients.^{21, 23} We did not have information about serum IgG subclass on all patients in this study, and are therefore unable to comment on this aspect of immunoglobulin deficiency in CLL.

In summary, our study indicates that ~25% of newly diagnosed CLL patients have hypogammaglobulinemia and an additional ~25% with normal immunoglobulin levels at diagnosis will develop hypogammaglobulinemia during the course of their disease. Hypogammaglobulinemia at the time of diagnosis appears to predict for shorter TFT, independent of other risk factors. Although the association between hypogammaglobulinemia and risk of infection is well documented, hypogammaglobulinemia does not appear to impact overall survival.