RTI: Respiratory Tract Infection; SmPC: Summary of Product Characteristics; AE: Adverse Event; ADR: Adverse Drug Reaction; C-M-H: Cochran Mantel Haenszel; COPD: Chronic Obstructive Pulmonary Disease; SADR: Serious Adverse Drug Reaction; SAE: Serious Adverse Event; CIOMS: Council for International Organizations of Medical Sciences; ERS: Erythrocyte Sedimentation Rate; ASAT: Aspartate Aminotransferase; ALAT: Alanine Aminotransferase; Plc: Placebo; tid: 3 times daily

Respiratory tract infections (RTI) are among the most frequent of all infectious conditions. Taking into consideration that most RTI are of viral origin, the inappropriate use of antibiotics for initial treatment has become an increasingly relevant public health concern because of potential resistance development [1]. Unless an acute tonsillitis accompanied by a group A-β haemolysing streptococci infection or bacterial pneumonia is present, there may not be any need for antibiotic treatment at all [2-4]. Herbal medicines are frequently used to treat RTI; in many cases, however, convincing clinical trial data is lacking. On the other hand, increasing interest in these products on part of patients and physicians underscores the need for such compounds.

EPs 7630^* is a herbal drug preparation from the roots of Pelargonium sidoides (1:8-10), extraction solvent ethanol 11% [w/w], widely used in Europe, the Commonwealth of Independent States, and Latin America for the treatment of RTI. Several in-vitro-evaluations of EPs 7630 and its isolated constituents have demonstrated pharmacological activities including moderate direct antibacterial and antiviral action and notable immune-modulatory capabilities: immune-modulatory activities are mediated mainly by the release of tumour necrosis factor α and nitric oxides, the stimulation of interferon β, and an increase in natural killer cell activity [5-8]; further biological activities in vitro are improved phagocytosis, oxidative burst and intracellular killing by human peripheral blood phagocytes, and inhibition of interaction between group-A-streptococci and host epithelia [9,10].

Several open-label and randomised double-blind clinical studies have been published over the last years in the indications of tonsillopharyngitis, bronchitis and sinusitis [11-30]. While systematic reviews and meta-analyses of EPs 7630 for acute bronchitis and RTI [31,32] have already shown encouraging evidence from currently available data that EPs 7630 is effective in patients with acute bronchitis, the complete evidence available on clinical tolerability and safety of EPs 7630 has not been presented so far. Recently, a first step to solve this issue was undertaken by Teschke et al. [33,34] who performed in-depth re-evaluations of 32 reports of purported Pelargonium sidoides liver function alterations available to the German authorities going back to the year 2004. It was shown that a probable or highly probable causal relationship with Pelargonium sidoides could not be assumed in any of the cases, and the authors concluded that convincing evidence is lacking that the herbal drug was a potential hepatotoxin in the analysed cases [33,34].

This review adds to these findings by providing the first overview of safety and tolerability of EPs 7630 based on publications and study reports from clinical trials and post-marketing surveillance studies. According to the current guideline on Summary of Product Characteristics (SmPC) the frequency of adverse reactions should be derived from pooled placebo-controlled studies [35], the review focuses on placebo-controlled clinical pivotal and pilot trials. Available data derived from non-placebo-controlled trials and interaction studies with EPs 7630 are additionally discussed.

Data Sources

Data reviewed were extracted from publications and unpublished reports of 29 studies investigating the efficacy and tolerability of EPs 7630. They represent all clinical research projects involving EPs 7630 that were funded by the manufacturer and completed by February 2010. Clinical study reports of unpublished trials were provided by Dr. Willmar Schwabe GmbH &Co. KG, Karlsruhe, Germany; they also include pilot studies and hypothesis-generating trials.

Analyses

Incidence rates of adverse events (AEs) were determined separately for each primary study as well as across the studies in an integrated analysis, based on pooled raw data from the primary trials.

In a first step, AEs observed during the analysed studies were classified according to MedDRA Preferred Terms. AEs that were in line with those listed in the SmPC of the marketed product were then assigned to system groups in a second step which reflect adverse drug reactions that may occur seldom (i. e., in 1–10 patients out of 10,000 exposed) or occasionally (i.e., in 1–10 patients out of 1,000 exposed) during treatment with EPs 7630 (gastrointestinal complaints, hypersensitivity reactions, nasal bleeding, gingival bleeding, and liver associated events). An integrated analysis of AEs assigned to the 5 system groups was performed.

Independently of the outcome seen for the integrated analysis, additional analyses of liver enzyme parameters for changes over time were performed, which included ALAT, ASAT, γ GT and serum bilirubin (total, direct, and indirect), in order to further investigate potential hepatotoxic effects of EPs 7630.

The integrated analysis of AEs assigned to the 5 system groups was performed by pooling the data from all 19 placebo-controlled trials available for review. In studies with more than one EPs 7630 dosage level, all patients receiving the herbal drug were pooled. For event rates within the treatment groups, 95% confidence intervals were determined using the Wilson’s method (Newcombe method 3) [36]. To compare the relative frequencies of patients with AEs and adverse drug reactions (ADRs) between EPs 7630 and placebo, event risk differences with their 95% confidence intervals were calculated for each system group according to the Newcombe Method 10 [37] and the Cochran-Mantel- Haenszel (C-M-H) test for risk differences between EPs 7630 and placebo were performed. The calculated p-values are to be interpreted in a descriptive manner as no formal hypotheses were prespecified and no adjustment for multiplicity was applied.

Study Designs, Dose Regimens, Number of Patients

Our review includes 19 double-blind placebo-controlled therapeutic studies (Table 1), 1 double-blind placebo-controlled pharmacokinetic interaction study in healthy volunteers, 1 double-blind active-controlled study, 2 open-label active-controlled studies and 6 non-controlled open-label studies. In studies performed in patients the therapeutic indications were acute tonsillopharyngitis, acute sinusitis, acute bronchitis, chronic obstructive pulmonary disease (COPD), acute and chronic airway infections, and acute rhinopharyngitis. Eleven studies enrolled only children, 14studies only adults, and 4 studies enrolled both children and adults.

Table 1: Overview of safety and tolerability from placebo-controlled studies.

Across all studies a total of 10,026 patients and 31 healthy subjects were included. Of these, 8,005 individuals were exposed to EPs 7630, 1,883 to placebo, 139 to the comparator acetylcysteine (bronchitis), and 30 to symptomatic therapy not otherwise specified (indication: tonsillopharyngitis). In total, 3,939/10,026 patients (39.3%) were infants, children, or adolescents up to the age of 18 years, of whom 3,243 were exposed to EPs 7630 and 527 to placebo.

In adults and children above 12 years of age, EPs 7630 dose regimens ranged from 30 drops (tonsillopharyngitis, sinusitis, bronchitis, rhinopharyngitis, COPD) to 60 drops thrice daily (sinusitis). In 1 study, 30 drops every hour up to a maximum of 12 times daily for the first 2 days followed by 30 drops thrice daily were administered to adults with sinusitis.

In children ≤ 12 years, EPs 7630 dose regimens included 5 or 10 drops (bronchitis), 10 to 20 drops (tonsillopharyngitis), and 20 drops thrice daily (sinusitis, bronchitis). In 4 studies, 20 drops every hour up to a maximum of 12 times daily for the first 2 days followed by 20 drops thrice daily were administered for the indications tonsillopharyngitis, sinusitis and bronchitis. In 2dose-finding studies, patients received film-coated tablets with 10 mg, 20 mg or 30 mg EPs 7630thrice daily.

Overall treatment duration depended on the respective indications and ranged from 6 to 10 days in tonsillopharyngitis (up to 5 weeks in case of recurrent tonsillopharyngitis), 7 to 14 days in bronchitis, 10 days in rhinopharyngitis, 3 to 4 weeks in sinusitis (up to 12 weeks in case of chronic recurrent sinusitis) and 24 weeks in COPD.

Safety and tolerability

No serious adverse drug reactions (SADR) were reported in any of the studies during the treatment period. Overall, 4serious adverse events (SAEs) were documented (EPs 7630: 3, placebo: 1). All SAEs were assessed as being unrelated to the investigational medication since all events concerned independent infectious diseases (infectious enteritis in 2 cases, 1 case each of diphtheria and pneumonia).

Data from placebo-controlled trials

Overview of individual adverse event and premature withdrawal rates: Table 1 summarises AE information from the 19 double-blind, placebo-controlled trials.

The percentage of patients reporting any AEs (independently of the associated causality assessment) ranged between 1.4% and 51.5% for EPs 7630 and between 0.0% and 39.6% for placebo. In 7 out of the 19 placebo controlled trials both treatment groups showed AE rates of <10% and in 10 trials AE rates of <20% were observed for both groups. AE rates ≥ 25% in any treatment group were observed in studies no. 3, 5, 12, 16, and 19. All studies with AE rates of ≥ 25% in one group were among the ‘smaller’ trials included in this review, with total sample sizes between 51 and 200 patients. Study no. 19, which exhibited the highest AE rates in both treatment groups, was a long-term trial in COPD, with treatment duration of 24 weeks, whereas all other trials investigated acute conditions and had a period of observation of 3 weeks or less. Due to the substantially longer period of observation the participants of study no. 19 were thus at a higher risk of experiencing at least one AE. As shown in Table 1, the majority of the most frequently reported AEs in double-blind, placebo-controlled trials were among the events included in the SmPC of EPs 7630.

Rates of AE related withdrawals from treatment of 1% or less were observed in 9 out of the 19placebo-controlled studies for EPs 7630 and in 10 studies for placebo. Fourteen and 12 studies showed withdrawal rates ≤ 2% for EPs 7630 and placebo, respectively. AE related withdrawal rates ≥ 10% in any treatment group were observed in 5 studies (no. 1-5). Out of 14 placebo-controlled trials with any AE-related withdrawals 9 showed lower withdrawal rates in the EPs 7630 group.

Integrated analysis of adverse events across all placebocontrolled trials according to the 5 system groups: Results of the integrated analysis of AEs and ADRs in the 19 placebo-controlled trials focusing on the 5system groups are summarized in Table 2. Data from 4,345 patients (EPs 7630: 2,478; placebo: 1,867) were included in this analysis.

Table 2: Incidence of adverse events based on pooled data from double-blind, placebo-controlled trials – number (%) of patients and 95% confidence intervals.

For AEs with any causal relationship to the investigational treatments, significant differences between EPs 7630 and placebo at a descriptive p value level of 5% were detected for AEs in the system groups gastrointestinal complaints (6.0% and 3.1% for EPs 7630 and placebo, respectively; two-sided C-M-H-test p-value: p<0.001), hypersensitivity reactions (0.7% and 0.3%; p=0.036) and epistaxis (1.2% and 0.5%; p=0.028). Further differences in AE rates between placebo and EPs 7630 occurred at the 10% p value level for gingival bleeding (0.2% and 0.0% for EPs 7630 and placebo, respectively; p=0.071).

Regarding potential ADRs, interpretable differences between placebo and EPs 7630 could only be observed for events related to the system groups gastrointestinal complaints (3.6% and 1.8% for EPs 7630 and placebo, respectively; p=0.002) and hypersensitivity reactions (0.4% and 0.1% for EPs 7630 and placebo, respectively; p=0.036). Comparisons in the other system groups revealed no statistically relevant differences between the treatments.

This also holds true for the system group concerning hepatobiliary events: Liver associated AEs (any causality assessment) were reported under EPs 7630 or placebo in 6 out of the 19 placebo-controlled trials. In 4 trials such events were more frequent in patients exposed to EPs 7630 and in 2 trials higher event rates were observed in the placebo group. Across all 19 trials the rates of hepatobiliary AEs were 0.48% for both EPs 7630 and placebo, with a risk difference of 0.00% (95% confidence interval: [0.48%; 0.43%]; p=0.517). Across all trials, 12 patients exposed to EPs 7630 and 9 patients treated with placebo were affected. Potentially treatment related hepatobiliary AEs were observed in 3 out of the 19 double-blind, placebo-controlled trials, with over-all event rates of 0.20% and 0.11% for EPs 7630 and placebo, respectively (risk difference: 0.09%; 95% confidence interval: [-0.21%; 0.38%]; p=0.266). In 6 out of the 12 hepatobiliary AEs in the EPs 7630 group and in 5 out of the 9 events in the placebo group the patients had abnormal liver enzyme values already at baseline.

Influence on liver enzyme parameters: Although the integrated analysis of adverse events across all placebo-controlled trials revealed no statistically relevant treatment group differences for the system group concerning hepatobiliary events, safety laboratory values were additionally analysed in order to further investigate the purported hepatotoxicity of the herbal drug. ASAT, ALAT, and γ GT were determined in 11 out of the 19 double-blind, placebo-controlled trials (representing a total of 1,578 patients exposed to EPs 7630 and with valid post-treatment values) whereas serum bilirubin was assessed in 3 trials (with 197 evaluable patients treated with EPs 7630 for analysis of total bilirubin and 136 patients for direct and indirect bilirubin). Table 3 shows the number and percentage of patients with values inside the applicable reference range at baseline and elevated values at treatment end, and vice versa. For each parameter the table also presents the average within-group difference between baseline and treatment end. For all parameters except indirect bilirubin, for which only a moderate number of evaluable patients were available, the percentage of study participants treated with EPs 7630 who had normal values at baseline but abnormal values at treatment end was on one level with placebo. Furthermore, the percentage of patients who shifted from normal to abnormal values was always substantially smaller than the percentage of those whose initially abnormal values returned to the reference range at treatment end.

Table 3: Liver enzyme values in double-blind, placebo-controlled trials (patients/total and % value, or mean ± SD and 95% confidence interval).

Under both EPs 7630 and placebo treatment, the changes between baseline and treatment end observed in the study groups’ mean values were close to zero, indicating that parameter increases and decreases were about evenly distributed among the patients.

Data from Non-Placebo-Controlled Studies and Interaction Studies: This section summarises AE information derived from a placebo-controlled interaction study with EPs 7630 and penicillin V in healthy subjects [38] and from9non-placebo-controlled therapeutic studies (3post-marketing surveillance studies [18,20,21,25,39], 3 openlabel outcomes studies [12,16,24], 2 reference-controlled studies versus acetylcysteine [17,40], and a study versus symptomatic treatment [14]), which enrolled a total of 5,527patients exposed to EPs 7630.

In the therapeutic studies the most frequently reported AEs were gastrointestinal symptoms, e.g. diarrhoea, nausea and epigastric pain, with incidences of up to 1.5% under EPs 7630. The AE rates in the penicillin interaction study in a total of 31 healthy volunteers generally exceeded those in the therapeutic studies; however, AE incidences in the placebo group were higher than in subjects randomised to EPs 7630 (most frequent events: gastrointestinal complaints 13.3% and 62.5%, headache 13.3% and 18.8% of subjects for EPs 7630 and placebo, respectively). No SADRs were reported in any of the studies presented in this section, and the percentage of patients who discontinued their study participation due to an AE did not exceed 3.9%.Furthermore, the results of the interaction study [38] did not indicate any interaction potential between EPs 7630 and penicillin.

Our analyses of 29 clinical trials and post-marketing surveillance studies involving a total of more than 8,000 patients exposed to EPs 7630 show that the herbal drug was well tolerated. The favourable safety profile of EPs 7630 is underlined by the fact that no SADRs were reported in this large patient population. AE rates under EPs 7630 were generally somewhat higher in randomised, controlled studies as compared to post-marketing surveillance and open outcomes studies, but this was to be expected which is in line with current literature surveys and reviews [41,42]. In the 19 double-blind, randomized, placebo-controlled trials covered by our review, AEs observed under EPs 7630 and placebo were similar with regard to type and incidence for the majority of system groups or symptoms. Although patients treated with the herbal drug exhibited a slightly higher over-all risk of AEs than those randomised to placebo, there was a higher percentage of AE related withdrawals in the placebo group.

The overview of the results of the placebo-controlled studies shows that the majority of the observed AEs reflect the system groups listed in the SmPC of the marketed products. EPs 7630 was associated with slightly higher incidences of epistaxis, gingival bleeding, hypersensitivity reactions, and gastrointestinal complaints (notably diarrhoea, nausea and epigastric pain). However, the only system group where the risk difference between EPs 7630 and placebo exceeded 1% was gastrointestinal complaints, and the only single event for which an increase by more than 0.5% above the placebo level was observed was epistaxis. These findings are in line with safety results obtained in post-marketing surveillance studies which were conducted in a larger patient population and therefore match most closely the general usage conditions of EPs 7630.

Recently there has been a discussion about suspected, rare hepatotoxic effects of preparations containing Pelargonium sidoides extract in the aftermath of the submission of spontaneous reports of hepatobiliary reactions (mostly inflammatory liver diseases and cholestasis) to the competent pharmacovigilance authorities in Germany [43,44]. Considering the fact that Pelargonium sidoides extract contains natural coumarin derivatives [45], potential hepatotoxic effects or bleeding complications might give rise to concerns.

However, hepatotoxicity of coumarin (1,2-benzopyrone) is attributed to metabolic activation to an epoxide intermediate, coumarin 3,4-epoxide. Whereas rats are most susceptible to coumarin-induced hepatotoxicity, in most species, particularly in humans, coumarin is predominantly hydroxylated to 7-hydroxycoumarin, a nontoxic metabolite [46-48]. Moreover, coumarin and 7-hydroxycoumarin have been shown to possess antitumour and immunomodulatory effects [49,50]. Since EPs 7630 exclusively contains 7-hydroxycoumarin derivatives, it is not surprising that no pharmacological evidence for hepatotoxicity has been obtained [51]. Likewise, in animal experiments, an influence of EPs 7630 on plasma coagulation, as well as possible pharmacokinetic or pharmacodynamic interactions with warfarin, could be excluded [45]. Moreover, as the 7-hydroxycoumarin derivatives do not possess the structural characteristics required for anticoagulant activity, it appears unlikely that an increased tendency towards bleeding complications arises in patients due to intake of EPs 7630 [45].

Teschke and colleagues [33] performed an in-depth re-evaluation of all reports of purported Pelargonium sidoides hepatotoxicity submitted to the German authorities and identified confounding factors like multiple final diagnoses unrelated to Pelargonium sidoides as well as poor quality of data and a lack of basic diagnostic measures to exclude even common diseases of the liver in nearly all cases. They determined that according to the criteria proposed by the Council for International Organizations of Medical Sciences (CIOMS) [52] a probable or highly probable causal relationship with Pelargonium sidoides could not be assumed in any of the cases and concluded that convincing evidence is lacking that the herbal drug was a potential hepatotoxin in the analysed cases. The results were supported by an in-depth re-evaluation of 13 additional spontaneous reports of suspected hepatotoxicity from which the same conclusions were drawn [34].

The findings of Teschke and colleagues [33,34] are in line with the results of the analyses of data from clinical trials and post-marketing surveillance studies presented in this paper. These analyses revealed neither an incidence of hepatobiliary adverse events in patients exposed to EPs 7630 above the level observed in patients treated with placebo, nor a treatment associated increase of abnormal liver enzyme values or a shift in these safety laboratory parameters’ mean values.

The results are also consistent with recently published studies involving the administration of EPs 7630 but not included in our review [53-56] for which no adverse effects on the hepatobiliary system were reported. Current evidence therefore does not provide any evidence that EPs 7630 could have a hepatotoxic effect.

EPs 7630 is a well-tolerated herbal medicine in the management of RTI in children and adults. Evidence for hepatotoxic effects in humans during routine administration was neither provided in the literature, nor by our own analyses.

We thank Dr. Willmar Schwabe GmbH & Co. KG, Karlsruhe, Germany, for allocation of funds, clinical data, and manuscripts of the described studies.

HM and WK were reimbursed for their scientific services for this review on a contractual basis by Dr. Willmar Schwabe GmbH & Co. KG, Karlsruhe, Germany. SK is an employee of Dr. Willmar Schwabe GmbH & Co. KG, Karlsruhe, Germany, and serves as Qualified Person for Pharmacovigilance for the company.

^*EPs® 7630 is the active ingredient in Umckaloabo® (ISO-Arzneimittel, Ettlingen, Germany)