Impact of screening and exclusion of high anti-A titer donors on the risk of hemolytic anemia with intravenous immunoglobulin treatment: A hospital-based cohort study in the US.

Hemolytic anemia is a potential complication of administration of intravenous immunoglobulin preparations (IVIG).1 The hypothesized predominant mechanism is the passive acquisition of A and B isoagglutinins (anti‐A and anti‐B antibodies) from the IVIG product,2 related to the antibody quantity per body weight. Hemolytic anemia is more common in patients receiving high dose IVIG for immune modulation than for low dose immunoglobulin replacement therapy in immunodeficiency.

IVIG products are derived from large human plasma pools of many individual donations, containing immunoglobulin G class isoagglutinins that may co‐purify with other antibodies in the IVIG production process. A isoagglutinins titers are higher than B isoagglutinin titers in the donor population and A patients are more frequent than B patients.3

Recent reports of increased hemolytic anemia rates relate to the change in the IVIG production from Cohn's method (cold‐ethanol fractionation), that reduced isoagglutinins, to chromatographic processes with higher purity and functional integrity.4 One available IVIG product is Privigen (CSL Behring, Bern, Switzerland) that until 2013 was produced with a chromatographic process lacking dedicated steps to reduce isoagglutinins. To reduce the isoagglutinin titers and the risk of hemolytic anemia with Privigen, a two‐ step approach was implemented: exclusion of donors with high anti‐A titer donors introduced in mid‐2013,4 followed by an immunoaffinity chromatography step which started in Oct‐2015.5

We conducted a study to describe the risk of hemolytic anemia before and after the first step of exclusion of high anti‐A titers donors from pooled plasma in the manufacturing of the IVIG Privigen.

Data were obtained from the United States (US) Premier Perspective database. We analyzed data from 862 US hospitals of patients treated with Privigen intravenously. Two study cohorts were formed of patients with use of Privigen before and after excluding high titer anti‐A donors: Jan‐2008 through Dec‐2012 (period 1, Privigen produced without isoagglutinin reduction) and Oct‐2013 through Dec‐2015 (period 2, Privigen produced mainly from anti‐A screened donors).

Exposure of interest was treatment with Privigen, determined for each patient and study cohort. Treatment episodes were defined as units of continuous duration of IVIG. Patients in each cohort were observed from the first Privigen administration until the earliest of: 30 days after the last Privigen administration, the occurrence of first hemolytic anemia, end of the respective period, or death. The cumulative IVIG dose per kilogram (kg) body weight and treatment episode was estimated from the daily quantity of Privigen administered and the corresponding median age‐ and sex specific body weight from the US population.6

The primary outcome was hemolytic anemia within 30 days of IVIG administration assessed from manual review of patient summaries by a hematologist (TLS) for the assessment of all potential hemolytic anemia events, the date of hemolytic anemia and the indication for IVIG use.

Indications for Privigen use were history of immunodeficiency, malignant neoplasm of lymphatic and hematopoietic tissue, immune thrombocytopenia, Kawasaki disease, chronic inflammatory demyelinating polyneuropathy, Guillain‐Barré syndrome, and myasthenia gravis.

Crude incidence rates of hemolytic anemia temporally related to Privigen use were calculated for each period and stratified by treatment setting (inpatient or outpatient), age, sex, Privigen indication, and Privigen dose per kg body weight. Incidence rates of hemolytic anemia in period 2 were compared with respective rates in period 1 by calculating incidence rate ratios (IRR) from Poisson regression, using the incidence rate in period 1 as reference and adjusted for treatment setting, sex, age, Privigen indication and dose per body weight. The study protocol was approved by the European Medicines Agency, protocol number ENCePP/SDPP/6040.

The cohorts in period 1 and 2 consisted of 9099 and 7431 Privigen users; mean age, 50.7 and 46.5 years respectively. Common indications were immune thrombocytopenia, immunodeficiency and malignant neoplasm.

The hemolytic anemia rate was 1.05/10 000 person‐days (95% CI, 0.82‐1.34) in period 1 and 0.82 (0.58‐1.12) in period 2, adjusted IRR 0.82 (0.55‐1.23) for period 2 vs. period 1 (P = .17).

The incidence rate of hemolytic anemia per 10 000 person‐days was higher in period 1 compared with period 2 in patients with immune thrombocytopenia [3.92 (2.70‐5.51) versus 1.78 (0.85‐3.28)], in patients with Kawasaki disease [5.43 (1.48‐13.89) versus 0.0 (0.00–6.01)], and in patients with a total Privigen dose ≥1.75 g/kg body weight per episode [3.41 (2.18‐5.07) versus 1.44 (0.66‐2.73)] (Table 1). A marked increase in the rate of hemolytic anemia with Privigen dose seen in period 1 (0.49 at low dose to 3.41 at high dose, P < .01), was attenuated in period 2 (0.53 to 1.44, P = .21).

Table 1

Incidence rates of hemolytic anemia in temporal association with Privigen use in period 1 and period 2, overall and stratified by Privigen indication and dose

	Period 1Jan 2008 to Dec 2012			Period 2Oct 2013 to Dec 2015			Comparisona
	HA cases	Person‐days at risk	Crude IRb (95% CI)	HA cases	Person‐days at risk	Crude IRb (95% CI)	Crude IRRc (95% CI)	Adjusted IRRc, d (95% CI)	P‐valuec, e
Total	68	644756	1.05 (0.82‐1.34)	39	476931	0.82 (0.58‐1.12)	0.78 (0.52‐1.15)	0.82 (0.55‐1.23)	.17
Privigen indicationf
Immune thrombocytopenia	33	84083	3.92 (2.70‐5.51)	10	56129	1.78 (0.85‐3.28)	0.45 (0.22‐0.92)	0.46 (0.22‐0.93)	.02
Immunodeficiency	15	380985	0.39 (0.22‐0.65)	8	266318	0.30 (0.13‐0.59)	0.76 (0.32‐1.80)	0.69 (0.29‐1.66)	.21
Malignant neoplasm	16	117912	1.36 (0.78‐2.20)	7	77042	0.91 (0.37‐1.87)	0.67 (0.28‐1.63)	0.69 (0.28‐1.70)	.21
GBS	5	20292	2.46 (0.80‐5.75)	1	18863	0.53 (0.01–2.95)	0.22 (0.03‐1.84)	0.25 (0.03‐2.25)	.11
Kawasaki disease	4	7373	5.43 (1.48‐13.9)	0	6135	0.00 (0.00–6.01)	‐	‐	‐
Myasthenia gravis	1	36352	0.28 (0.01‐1.53)	4	34941	1.14 (0.31‐2.93)	4.16 (0.47‐37.3)	4.99 (0.55‐45.5)	.92
CIDP	1	53534	0.19 (0.00–1.04)	1	53020	0.19 (0.00–1.05)	1.01 (0.06–16.2)	1.12 (0.07–18.7)	.53
Other or unknown indication	9	70141	1.28 (0.59‐2.44)	14	50857	2.75 (1.50‐4.62)	2.15 (0.93‐4.96)	2.32 (1.00–5.40)	.98
Privigen dose (g/kg body weight)
<0.75	20	410763	0.49 (0.30‐0.75)	15	285409	0.53 (0.29‐0.87)	1.08 (0.55‐2.11)	1.07 (0.54‐2.11)	.58
≥0.75 to <1.75	19	111137	1.71 (1.03‐2.67)	13	110991	1.17 (0.62‐2.00)	0.69 (0.34‐1.39)	0.90 (0.44‐1.84)	.39
≥1.75	24	70467	3.41 (2.18‐5.07)	9	62565	1.44 (0.66‐2.73)	0.42 (0.20‐0.91)	0.48 (0.22‐1.04)	.03
Unknown dose	5	52389	0.95 (0.31‐2.23)	2	17966	1.11 (0.13‐4.02)	1.17 (0.23‐6.01)	1.09 (0.20‐5.88)	.54

CI: Confidence interval; CIDP: Chronic inflammatory demyelinating polyneuropathy; GBS: Guillain‐Barré Syndrome; HA: Hemolytic anemia; IR: Incidence rate; IRR: Incidence rate ratio.

Using period 1 as reference.

IR per 10 000 person‐days at risk.

Omitted when no case detected in period 1 or 2.

Adjusting for treatment setting, sex, age, Privigen indication and Privigen dose using Poisson regression. Overall adjusted IRR estimate was adjusted for all covariates (i.e., treatment setting, sex, age, Privigen indication and total Privigen dose). All other IRR estimates were adjusted for all covariates except for the respective stratifying covariate.

Using one‐sided Wald test for multivariate Poisson regression.

More than one indication per Privigen episode possible.

With Privigen administered for immune thrombocytopenia and period 1 as the reference, the adjusted IRR was 0.46 (P = .02), and when given at high dose (≥1.75 g/kg body weight) it was 0.48 (P = .03, Table 1). The relative risk reduction of 52% resulted in an absolute reduction of 1.55 hemolytic anemia cases avoided per 10 000 person‐days of high‐dose IVIG treatment in period 2 vs 1.

In summary, using a large hospital‐based cohort of over 16 000 patients treated with Privigen we found a decrease in the risk of hemolytic anemia in association with exclusion of donors with high anti‐A titers from plasma pools used in manufacturing IVIG. This association was dose‐dependent and observed in indications requiring higher IVIG doses. In indications requiring ≥1.75 g IVIG/kg body weight there was a significant risk reduction of 52% associated with exclusion of high titer anti‐A donors (P = .03).

This study supports an association between dose of IVIG treatment and risk of hemolytic anemia and indicates that the exclusion of donors with high titers of anti‐A from plasma pools was associated with a reduction of risk with high dose therapy.