Angioedema due to acquired C1-inhibitor deficiency: spectrum and treatment with C1-inhibitor concentrate

Background Acquired angioedema due to C1-inhibitor (C1-INH) deficiency (AAE-C1-INH) is a serious condition that may result in life-threatening asphyxiation due to laryngeal edema. It is associated with malignant B-cell lymphoma and other disorders. The purpose of this study was to describe the characteristics and associated disorders of patients with AAE-C1-INH and assess the efficacy of plasma-derived C1-INH concentrate (pdC1-INH) in the treatment of AAE-C1-INH. Forty-four patients with AAE-C1-INH from the Angioedema Outpatient Service of Mainz were assessed for associated disorders. In 32 of these patients, the duration of swelling attacks was measured before and after treatment with pdC1-INH (Berinert® (CSL Behring, Marburg, Germany)). The time between injection and complete resolution of symptoms and treatment effectiveness was provided by the patients. Results The following underlying disorders were present: monoclonal gammopathy of undetermined significance (47.7%), non-Hodgkin lymphoma (27.3%), anti-C1-INH autoantibodies alone (11.4%), and other conditions (4.5%). In 9.1% patients, no associated disorder could be found. AAE-C1-INH led to the detection of lymphoma in 75% of patients with the malignancy. Treatment with pdC1-INH shortened attacks by an average (SD) 54.4 (± 32.8) hours (P < 0.0001). The earlier the attack was treated, the shorter the time between injection and resolution of symptoms (P = 0.0149). A total of 3553 (97.7%) of the 3636 attacks were effectively treated with pdC1-INH as assessed by the patient. The mean (SD) dose per-attack was 787 (± 442) U. pdC1-INH was effective in 1246 (93.8%) of 1329 attacks in 8 patients with anti-C1-INH autoantibodies and in 344 (99.4%) of 346 attacks in 6 patients without autoantibodies. The average (SD) dose per effectively treated attack was 1238.4 (± 578.2) U in patients with anti-C1-INH autoantibodies and 510.2 (± 69.1) U in patients without autoantibodies. Conclusions pdC1-INH is highly effective in treating AAE-C1-INH patients and is also effective in the vast majority of attacks in patients with anti-C1-INH autoantibodies. It is fast-acting and reduces attack duration.

Diagnosis of AAE-C1-INH generally occurs after 40 years of age [11]. There is no genetic association or family history of angioedema in AAE-C1-INH; this is in contrast to the types of hereditary angioedema (HAE) which are caused by or associated with specific mutations in the genes coding for C1-INH, factor XII, plasminogen or angiopoietin [12][13][14][15]. Patients with AAE-C1-INH have low plasma levels of C1-INH (functional and antigenic) and C4 which are usually below 50% of normal. Unlike HAE-C1-INH patients, the majority of AAE-C1-INH patients also have reduced C1q levels. The presence of anti-C1-INH autoantibodies and the absence of C1-INH genetic mutations can aid in the diagnosis of AAE-C1-INH [4].
Currently, there is no approved treatment for AAE--C1-INH. HAE-C1-INH therapies such as plasma-derived C1-INH concentrate (pdC1-INH) and the bradykinin B2 receptor antagonist, icatibant, are used to resolve AAE-C1-INH attacks [16]. It is generally supposed that these on-demand treatments are most effective when administered early in the attack [17]. However, published information on the efficacy and safety of AAE-C1-INH treatments is limited and has not been studied systematically.
Therefore, the aim of this observational, retrospective study was to: (1) describe the characteristics and associated disorders of patients with AAE-C1-INH, (2) assess the efficacy of pdC1-INH in the treatment of AAE-C1-INH patients (3) assess the effect of time to injection, dose and anti-C1-INH autoantibodies on pdC1-INH efficacy.

Patient characteristics
Patients with a confirmed diagnosis of AAE-C1-INH attending our outpatient clinic, Department of Dermatology, University of Mainz, Germany (AOSM), were eligible for this observational study. In total, 44 patients with AAE-C1-INH were identified and their health records were reviewed and analyzed. The prevalence of AAE-C1-INH compared with that of HAE-C1-INH in the AOSM was 1:9.3. The majority (61%) of patients were female and mean (SD) age at onset of recurrent angioedema of AAE-C1-INH was 56.2 (± 14.8) years (Table 1). Five (11.4%) of the 44 patients were aged under 40 years when the recurrent angioedema started. Two female patients had onset at 21 and 30 years and had anti-C1-INH autoantibodies; another male and female patient had onset at 37 and 39 years, respectively and had MGUS. The fifth patient had no associated disorders. In these 5 patients, molecular genetic testing for a mutation in SERPING1 was negative.
Clinical symptoms were abdominal attacks, skin swellings of the face, extremities and genitals as well as isolated tongue swellings and laryngeal attacks. An erythema marginatum preceded the symptoms in 2 patients. Seven patients died during the course of the study; 2 deaths were from underlying diseases. There were no angioedema-related deaths in this group.
(5) The 4 (9.1%) remaining patients had no associated disorders such as MGUS, non-Hodgkin lymphoma, anti-C1-INH autoantibodies, cancer or liver failure. Three of the 4 patients had low C1-INH activity and protein, low C4 and low C1q in plasma. One of the 4 patients, a female, had recurrent angioedema of the extremities and lips and abdominal attacks for 17 years (from age 25 to 42 years). During this time, the patient had low C1-INH function and protein and low C4 but normal C1q. For the last 4 years, the patient has had no clinical symptoms and all 6 tests for C1-INH function and protein and C4 revealed normal results. Mutations In 18 patients, recurrent angioedema was present more than 1 year before an associated disorder was diagnosed, in 23 patients both were simultaneously diagnosed and in 3 patients, recurrent angioedema developed later than the associated disorder. In 9 (75.0%) of 12 patients, the associated malignant lymphoma was detected following diagnosis of AAE-C1-INH. In the other 3 patients, the underlying lymphoma was already known before the first visit at the AOSM.

Plasma complement
All 44 patients had functional and antigenic C1-INH plasma levels which were below the normal range (Table 1). C4 levels were also below the normal range in all patients; C1q levels were below normal levels in 39 (88.6%) patients.

Efficacy of pdC1-INH concentrate treatment
The mean (SD) duration of untreated attacks was 89.9 (± 14.8) hours over all attacks in the 32 patients and 84 (± 31.8) hours on a per-patient basis (mean of 32 averages). The mean (SD) duration of the treated attacks was 27.9 (± 12.9) hours over all attacks in the 32 patients and 29.6 (± 16.4) hours on a per-patient basis. (Fig. 2). Treatment with pdC1-INH concentrate shortened attacks by an average of 54.4 (± 32.8) hours (confidence interval (CI): 42.5, 66.2) (P < 0.0001) on a per-patient basis (i.e. by 64.8%).
All 32 patients responded to treatment, with 12 (37.5%) patients experiencing relief in 30 min or less. The average time to first relief after injection was 1.0 (± 3.3) hour in all 32 patients and 3.6 h (± 8.2) on a per-patient basis. Resolution of symptoms occurred on average 26.4 (± 12.6) hours after injection in all 32 patients and 26.7 (± 15.0) hours on a per-patient basis.
A total of 3553 (97.7%) of the 3636 attacks were effectively treated as assessed by the patient. Eighty-eight (2.3%) attacks in 8 patients did not or did not sufficiently respond to pdC1-INH treatment. Among the 8 non-responders were 4 patients with anti-C1-INH autoantibodies (described further below).

Effect of time to injection on efficacy
The mean (SD) time between attack onset and injection was 1.5 (± 0.9) hours for the 32 patients and 3.0 (± 2.7) hours on a per-patient basis. Regression analysis showed that there was a linear association between the average time to complete resolution of symptoms and the average time to injection i.e. the earlier the attack was treated, the shorter the time between injection and resolution of symptoms (P = 0.0149). Sensitivity analysis, omitting 2 patients with extreme times to injection, confirmed the association (P = 0.0036). Similar regression analysis showed no clear association between the average shortening of attack duration and the time to injection (P = 0.0745).

Effect of dose on efficacy
In order to analyze the effect of dose, a subgroup analysis of patients using either 500 U or ≥ 1000 U in at least 90% of their acute injections was performed (Table 3). Overall, the mean (SD) dose per-attack was 787 (± 442) U. The majority of patients received either 500 U (n = 13) or 1000 U (n = 9) in more than 90% of attacks; 1 patient received 1500 U and 1 patient received 2000 U. Patients in both dose subgroups had similar body weights and mean times between attack onset and injection (Table 3).
In the 500 U dose group, attack duration was shortened by a mean (SD) 53.1 (± 25.9) hours and in the ≥1000 U dose group by a mean (SD) 58.8 (± 40.1) hours. In the 500 U dose group, the time between injection and resolution of symptoms was 27.6 (± 16.3) hours and in the ≥1000 U dose group it was 21.9 (± 11.3) hours. However, no association between dose and shortening of attack duration (P > 0.1) or dose and time between injection and resolution of symptoms (P > 0.1) could be shown by linear regression.

Effect of anti-C1-INH autoantibodies on efficacy
Eight AAE-C1-INH patients were positive for anti-C1-INH autoantibodies and 6 patients were negative. The remainder of patients were not tested for the presence of anti-C1-INH autoantibodies.
Attack duration was shortened by a mean (SD) 50.4 (± 38.2) hours in the 8 patients who tested positive for

Discussion
The clinical features of our patients diagnosed with AAE-C1-INH were similar to those described in previous AAE-C1-INH studies [1,4,18]. In our institution, we found that the incidence of the condition was 1 patient with AAE-C1-INH for every 9.3 patients with   [21,22].
In most patients in our series, onset of AAE-C1-INH occurred at age 40 years or older. However, there were a few patients who were less than 40 years when their AAE-C1-INH started. Among these were 3 patients with anti-C1-INH autoantibodies but no MGUS or malignant lymphoma. We conclude, therefore, that a diagnosis of AAE-C1-INH cannot be ruled out in patients less than 40 years old.
We found that AAE-C1-INH was associated with a variety of disorders including MGUS, malignant non-Hodgkin lymphoma, anti-C1INH autoantibodies, and other conditions. In some patients, there was no associated disorder identifiable. A total of 27.3% patients had an underlying lymphoma and in 75.0% of those patients the lymphoma was detected by monitoring the signs of AAE-C1-INH. This underscores the importance of early diagnosis of AAE-C1-INH. We also report that 25.0% of patients had a low-grade malignant lymphoma and approximately half of them had a splenic marginal cell lymphoma. This confirms that splenic marginal cell lymphoma is the most common type of underlying lymphoma in patients with AAE-C1-INH [23,24].     [25,26]. The younger age at onset of angioedema described here could be a special feature of this subgroup of patients. However, the small number of patients needs to be considered. At present, it is not clear whether lymphoproliferative disorders (MGUS and lymphoma) and anti-C1-INH autoantibodies have one common pathogenic mechanism leading to C1-INH deficiency.
Our results show that pdC1-INH reduces the average attack duration by more than 60%. It is highly effective in resolving attacks, as evaluated by attack and by patient analysis, most notably if treatment is administered early in an attack. According to patients' self-assessment, all patients responded well to pdC1-INH in nearly all of their attacks.
It has been reported that a few AAE-C1-INH patients have needed treatment with high doses of pdC1-INH or that some patients have become completely or partially resistant to this treatment [4]. Some of these patients had autoantibodies to C1-INH concentrate [1,27]. In contrast, our results show that the vast majority (93.8%) of attacks in patients with anti-C1-INH autoantibodies responded well to pdC1-INH and that the response rate was similar to patients without anti-C1-INH autoantibodies (99.4% of attacks). Therefore, we conclude that patients with anti-C1-INH antibodies can respond to pdC1-INH. However, in some patients with anti-C1-INH autoantibodies, doses to effectively treat attacks need to be more than double that in patients without autoantibodies. In our series, 4 out of 8 patients with anti-C1-INH autoantibodies did not respond sufficiently to pdC1-INH treatment in some attacks, whereas in other attacks in the same patients, treatment with pdC1-INH was effective. A variation in the level of anti-C1-INH autoantibodies may be a reason for this. Repeated and long-term investigations of the anti-C1-INH autoantibody levels in AAE-C1-INH patients are needed to explain the relationship between autoantibody level and treatment dose.
Given the low prevalence of the disorder, this is a relatively large observational study examining the clinical characteristics of AAE-C1-INH. The study recruited a considerable number of patients and recorded the details of a large number of attacks treated with pdC1-INH over a long period of time. The study is limited by its observational and retrospective study design involving data retrieval from patient reports, which may have resulted in some bias. The fact that not all patients could be tested for anti-C1-INH autoantibodies also limits the results.

Conclusions
In summary, our study demonstrated that the clinical symptoms of AAE-C1-INH are similar to those of HAE-C1-INH. Early diagnosis of AAE-C1-INH is important because of the risk of asphyxiation by laryngeal edema and because AAE-C1-INH is frequently associated with an underlying malignant disorder. AAE-C1-INH attacks can be treated with pdC1-INH, which is fast-acting and is highly effective in nearly all AAE-C1-INH attacks, including those in patients with anti-C1-INH autoantibodies.

Patients
Patients for this observational study were followed up at the AOSM from March 1986 to August 2017. Diagnosis of AAE-C1-INH was based on personal history of recurrent angioedema, no family history of angioedema and plasma examination of C1-INH, C4 and C1q. The absence of a C1-INH (SERPING1) genetic mutation confirmed diagnosis in some patients. The study was approved by the local ethics committee (Ethics Committee of the Landesärztekammer Rheinland-Pfalz, 837.413.13 (9098-F)) and all patients gave their informed consent to participate in the study.

Study design
The clinical health records of 44 patients with AAE-C1-INH were reviewed retrospectively and patient characteristics noted. Thirty-two of these patients were treated for acute attacks with pdC1-INH. The efficacy variables were defined as follows: (1) the duration by which attacks were shortened after treatment, (2) the time between injection and resolution of symptoms, and (3) patient assessed treatment efficacy. The duration of treated attacks was compared with the duration of untreated attacks in the same individuals. Untreated attacks were defined as those which occurred before the first administration of pdC1-INH. Treated attacks were defined as those treated with pdC1-INH. Attack duration was defined as the time between attack onset and resolution of symptoms. Data on the time to injection and the type and duration of the attacks was recorded using standardized questionnaires. The course of the attacks was divided into first relief from symptoms and complete resolution of symptoms. In addition, patients evaluated the treatment effectiveness themselves during interviews and/or documented it in a patient diary. The following treatment evaluations were recorded by patients: treatment effective (responders) and treatment not or not sufficiently effective (non-responders).

Treatment
The treatment group received intravenous, pasteurized pdC1-INH concentrate (Berinert® (CSL Behring, Marburg, Germany)) which was either self-injected or administered by the patient's general practitioner, at their local hospital or at our department. The usual dose comprised 500 U of pdC1-INH, which corresponds to a C1-INH plasma activity of about 500 mL of fresh plasma. For treatment of skin swellings, 500 U pdC1-INH was recommended. If the patient had a body weight over 80 kg or if the patient felt that the 500 U dose was not sufficiently effective, 1000 U pdC1-INH was administered. Patients were counseled that mild skin swellings should not be treated. Mild skin swellings were defined as swellings limited to the back of one hand or foot or swellings at one extremity or at the trunk with a diameter of < 20 cm. Treatment was recommended for all facial and genital swellings. In skin swellings of the extremities, treatment was only recommended when the swellings were > 20 cm in diameter or if the whole extremity was affected or when the swelling of one extremity was followed by a swelling of another extremity or parts of the trunk within 24 h. Doses higher than 1000 U were administered only if the clinical response was not sufficient. All patients received vaccination for hepatitis B virus.

Statistical analysis Patient characteristics
The following parameters were analyzed with univariate statistics (mean and SD) or frequency tables, as appropriate, for all patients and patients treated with pdC1-INH: baseline and demographic characteristics, associated disorders, monoclonal gammopathy antibodies, anti-C1-INH autoantibodies, prior and concomitant treatment, number of stays at hospital due to angioedema, number of stays in the intensive care unit, number of intubations, number of untreated swellings, and the average duration of untreated attacks. Univariate statistics for C1-INH function, C1-INH protein, C4 and C1q at first visit in AOSM were also generated for all patients and for patients treated with pdC1-INH.

Plasma-derived C1 inhibitor concentrate treatment
Univariate statistics for single doses of pdC1-INH were applied for both per-attack analysis and per-patient (on average doses) analysis. The number of attacks treated with pdC1-INH, the duration of pdC1-INH treatment, and the age at last pdC1-INH injection were analyzed with univariate statistics in a per-patient analysis. The number of attacks at different sites and the number of attacks treated with different doses of pdC1-INH were analyzed with a frequency table and frequency graph statistics in a per-attack analysis. In addition, a per-patient analysis was generated to show the number patients who preferably (i.e. in 90% or more of their attacks) received a special single dose of pdC1-INH.
For the time between onset of attack and injection, the time between injection and first sign of efficacy, the time between injection and complete resolution of symptoms, and the average shortening of attacks with treatment, univariate statistics were generated both in a per-patient and in a per-attack analysis.

Efficacy variables
The original dataset included data on the average duration of attacks with and without treatment with pdC1-INH for each patient as well as the number of attacks for each patient. The primary analysis on duration of attack was, therefore, a per-patient analysis which provided statistics on the average attack durations over all treated patients. In addition, a descriptive analysis on a per-attack basis was done by weighting the individual average duration of attacks by the individual number of attacks (untreated and treated).
Univariate statistics were applied for the average duration of attack treated with pdC1-INH and for the average duration of untreated attacks in both a per-patient and in a per-attack analysis. For the average shortening of the duration of attacks the following statistics were generated: univariate statistics, a two-sided Wilcoxon signed rank test (hypothesis: no shortening), and a 90% CI in a per-patient analysis.
In order to investigate the association between the time between injection and resolution of symptoms and the time between onset of attack and injection, linear regression analysis with graphical representation was generated. A similar regression analysis was performed for the association between average shortening of attacks and the time to injection, the association between shortening of attacks and the preferred single dose and the association between time to resolution of symptoms and the preferred single dose.
The percentage of effectively treated attacks, as assessed by the patient, was calculated for the total number of treated attacks, anti-C1-INH antibody positive patients, anti-C1-INH antibody positive patients with MGUS, and anti-C1-INH antibody negative patients. All analyses were performed using Excel or SAS.