To cite this article: Calderon M, Mösges R, Hellmich M, Demoly P. Towards evidence-based medicine in specific grass pollen immunotherapy. Allergy 2010; 65: 420–434.
When initiating grass pollen immunotherapy for seasonal allergic rhinoconjunctivitis, specialist physicians in many European countries must choose between modalities of differing pharmaceutical and regulatory status. We applied an evidence-based medicine (EBM) approach to commercially available subcutaneous and sublingual Gramineae grass pollen immunotherapies (SCIT and SLIT) by evaluating study design, populations, pollen seasons, treatment doses and durations, efficacy, quality of life, safety and compliance. After searching MEDLINE, Embase and the Cochrane Library up until January 2009, we identified 33 randomized, double-blind, placebo-controlled trials (including seven paediatric trials) with a total of 440 specific immunotherapy (SIT)-treated subjects in seven trials (0 paediatric) for SCIT with natural pollen extracts, 168 in three trials (0 paediatric) for SCIT with allergoids, 906 in 16 trials (five paediatric) for natural extract SLIT drops, 41 in two trials (one paediatric) for allergoid SLIT tablets and 1605 in five trials (two paediatric) for natural extract SLIT tablets. Trial design and quality varied significantly within and between SIT modalities. The multinational, rigorous trials of natural extract SLIT tablets correspond to a high level of evidence in adult and paediatric populations. The limited amount of published data on allergoids prevented us from judging the level of evidence for this modality.
We performed a systematic, evidence-based medicine (EBM) review of double-blind, placebo-controlled randomized clinical trials (DBPC RCTs) of commercially available grass pollen-specific immunotherapy (SIT) modalities for the treatment of seasonal allergic rhinoconjunctivitis (SAR) in adults and children (1). This health problem affects as many as one in seven of the population in Western countries (2, 3), with grass pollen as one of the major triggers (4).
Fundamental regulatory and technological changes are modifying the types of specific subcutaneous or sublingual SIT (SCIT and SLIT) product available to specialist practitioners. The European Medicines Agency (EMEA) and the United States Food and Drug Administration (FDA) have issued guidelines affecting the clinical development of SIT products (5, 6). With the exception of Gramineae grass pollen extract SLIT tablets (currently being registered across Europe as pharmaceutical specialties), SIT products are sold as either unregistered, named patient preparations or nationally registered formulations. Hence, in many European countries, the specialist physician initiating grass pollen SIT has to choose (in collaboration with the patient) between nonregistered and registered modalities. To inform that choice, we used an EBM approach to assess commercially available immunotherapy modalities.
Evidence-based medicine has been defined as ‘the conscientious, explicit and judicious use of current best evidence in making decisions about the care of the individual patient’ (7) and ‘the integration of the best available research evidence with clinical expertise and patient values’ (8). We evaluated the nature, quantity and quality of research evidence (essentially efficacy and safety) and patient values [quality of life (QoL) and treatment satisfaction] in DBPC RCTs. We did not seek to define the efficacy and safety of SIT itself or directly compare SCIT and SLIT, because these topics have been examined and discussed elsewhere (9–14).
Materials and methods
Search strategy and selection criteria
MEDLINE, Embase and the Cochrane Library were searched up until January 2009 using logical combinations of the following terms: rhiniti*; allerg*; seasonal*; rhinoconj*; hay fever; immunotherap*; immunolog*; densiti*; grass*; pollen*; pollinos*; Poaceae; Pooideae; Gramineae. We considered randomized DBPC RCTs of commercially available SCIT and SLIT modalities for grass pollen SAR and excluded studies considering safety only, mixtures of allergens and those looking primarily at asthma and perennial rhinitis. Articles were cross-checked against those identified in recent meta-analyses and reviews (9–14).
We extracted the following data (when reported) from each selected publication:
- 1 Study country/countries and year(s).
- 2 The SIT maintenance dose and the reported or calculated cumulative dose.
- 3 The randomized, intention-to-treat (ITT) and per-protocol (PP) populations in active treatment and placebo groups. Unless otherwise stated, the patient numbers quoted in this article refer to SIT-treated subjects (and not controls) in the ITT population.
- 4 Mean or median treatment duration.
- 5 Grass pollen count monitoring and the pollen season definition used.
- 6 The number of symptoms scored, the symptom scale (min/max), the total symptom score and variance for active and placebo groups and the P-value for any active vs placebo differences.
- 7 Rescue medications allowed, the medication scale (min/max), the overall medication score and variance for active and placebo groups and the P-value for any active vs placebo differences.
- 8 The patient’s QoL [on the Rhinoconjunctivitis Quality of Life Questionnaire (RQLQ)] and satisfaction with the treatment (15, 16).
- 9 Treatment-emergent adverse events (TEAEs), local and systemic adverse events (LAEs and SystAEs) and medication-driven serious adverse events (SAEs).
- 10 Treatment compliance.
We then grouped studies into (i) SCIT with natural antigen extracts, (ii) SCIT with modified antigens (‘allergoids’), (iii) SLIT with natural antigen extracts in a liquid formulation (drops), (iv) SLIT with natural antigen extracts in a tablet formulation and (v) SLIT with allergoids in a tablet formulation. We identified long-term studies (>12 months), in view of the EMEA’s recommendations on the need to demonstrate sustained efficacy and disease-modifying effects (5).
Thirty-three randomized DBPC RCTs met our selection criteria (25 studies in adults and eight studies in children, published from 1991 to 2009) (Fig. 1). They comprised a total of 3160 SIT-treated participants [SCIT with natural extracts: 440 subjects in seven trials (17–23); SCIT with allergoids: 168 SIT-treated subjects in three trials (24–26); SLIT with natural extract drops: 906 SIT-treated subjects in 16 trials (27–42); SLIT with allergoid tablets: 41 SIT-treated subjects in two trials (43, 44); SLIT with natural extract tablets: 1605 SIT-treated subjects in five trials (45–50)]. The main reasons for excluding studies were lack of double-blinding/randomization, review articles, multiple allergen extracts, outcomes unrelated to efficacy and preparations that are not commercially available. These criteria excluded early allergoid studies (51–55), a study with a recombinant allergen (which is not commercially available) (56), a paediatric SCIT study in which a combined asthma symptom-medication score was the primary outcome measure (57) and, regretfully, the ‘Preventive Allergy Treatment’ (PAT) study (58). Even though the 10-year PAT study was well designed and asked an important clinical question (does SCIT for allergic rhinoconjunctivitis prevent the development of asthma?), it failed to meet two of our study selection criteria. First, it was only double-blind for a small fraction of the study period. This is understandable, in view of the reluctance of ethics committees to authorize placebo injections in children over multiyear periods. Secondly, some polysensitized subjects were treated with a birch pollen extract and a grass pollen extract, making it difficult to draw conclusions on the clinical efficacy in grass pollen allergy. We also calculated an equally weighted Delphi trial quality score, as modified by Röder et al. (14, 59), and a Jadad trial quality score for the selected studies (60). However, given that all selected studies scored well (Delphi scores between 8 and 11 of a maximum of 11 and Jadad scores ranging from 2 to 5 of a maximum of 5), these scores were not discriminant from an EBM standpoint.
Even though we focused on a single allergen source, there was high between-study heterogeneity for the treatment effect on symptoms, as seen in recent meta-analyses (9–11). However, the SLIT natural allergen tablet modality (n = 7) studies or dose levels within a study) showed the lowest between-study variability [tau2 and I2 (95% confidence intervals) for Hedges’ g: 0.016 (58.5% (4.3–82%))], followed by SCIT with natural allergens (n = 7) with 0.039 [44.6% (0–76.7%)] and SLIT drops (n = 10) with 0.045 [51.9% (1.1–76.6%)]. The test was not applicable to SLIT with allergoid tablets and SCIT with allergoids, for which there were only one and two calculable studies, respectively. Meta-analysis and random effects meta-regression did not reveal a clear relationship between the treatment effect on one hand and SIT modality, dose or treatment duration on the other. Specifically designed clinical investigations are needed to address these precise questions, in the same way that optimal dosage has been determined in SLIT tablet trials.
|Authors, year of publication (countries), study year(s)||nR*||ITT (Ac/Pl)*||PP (Ac/Pl)*||Mean treatment time (months)||Symptoms: number measured [scale min–max], mean total symptom score Ac/Pl, P)||Medications: number recorded [scale min–max], mean score Ac/Pl, P)|
|SCIT with natural allergens|
|Varney et al., 1991 (UK), study year 1989||40||21/16||19/16||7||9 [0–27], NR, median AUC. Ac: 360; Pl: 928 (P = 0.001)||5 [0–7], NR, median AUC. Ac: 129; Pl: 627 (P = 0.002)|
|Zenner et al., 1997 (Germany), study year 1993||87||45/41||41/40||1.75||3 [0–21], AUC over 10 weeks: Ac: 82.2; Pl: 116.0 (P = 0.02)||4 [0–3], NR, mean medication days (out of 70), Ac: 26.6; Pl: 33.0 (P = 0.296)|
|Leynadier et al., 2001 (France), study year 1998||29||16/13||15/12||11||10 [0–30], AUC April 15–July 15: Ac: 49.6; Pl: 56 (P = NS)||6 [0 to NR], AUC April 15–July 15: Ac: 11.1 vs Pl: 40.8 (P = 0.005)|
|Frew et al., 2006 (UK), study year 2002||410||203 high Ac/104 low Ac/103 Pl||187 high Ac/89 low Ac/89 Pl||12||3 [0–9], mean nose + eye SS: high Ac: 3.13; low Ac: 3.44; Pl: 4.39 (P < 0.0001 for high Ac, P < 0.013 for low Ac)||5 [0 to NR], high Ac: 2.85; low Ac: 3.55; Pl: 4.21 (P = 0.0007 for high Ac, P = 0.16 for low Ac)|
|Dolz et al., 1996 (Spain), study years 1990–1992. Long-term study||28||18/10||NR||30||13 [0–39], NR, only Ac vs Pl difference stated: Season 1: ocular P < 0.001, nasal P = NS, lung P = 0.001||4 [0 to NR], NR. only Ac vs Pl differences stated: Season 1 (NS)|
|Durham et al. 1999, (UK), study years 1993–1995 (extension of Varney 1989 group). Long-term study||47||16 (AcM)/16 (AcD)/15 (Pl)||14 (AcM)/13 (AcD)/12 (Pl)||84 (AcM) and 48 (AcD)||9 [0–27]. NR. Median AUC over 11 weeks for Season 1: AcM = 626, AcD = 798; Pl: 2615 (P = NR)||4 [total 0 to NR]. NR. Median AUC over 11 weeks for Season 1: AcM = 756, AcD = 633; Pl: 3672 (P = NR)|
|Walker et al., 2001 (UK), study years 1996 (baseline), 1997–1998 (treatment). Long-term study||44||22/22||20/17||22||13 [0–39], median AUC for total SS over 11 weeks for Season 1: Ac: 525; Pl: 1160 (P = 0.08, NS)||5 [0 to NR]. NR. Median AUC over 11 weeks for Season 1: Ac: 308; Pl: 1127 (P = 0.1, NS)|
|SCIT with allergoids|
|Drachenberg et al., 2001 (Germany and Austria), study year 1999||141||81/60||74/50||1||2 [0–3] Ac: 2.03 (sum of 0.82 ± 0.58 and 1.21 ± 0.65); Pl: 2.58 (sum of 1.12 ± 0.52 and 1.46 ± 0.51) (P = 0.05)||5 [0 to unlimited]. Ac: 0.54 ± 0.71; Pl: 0.71 ± 0.77 (NS)|
|Brewczynski & Kroon, 1999 (Poland), study years 1991–1993. Long-term study||20||10/10||10/8||NR||2 [0–3]. AUC for calendar weeks 20–37 in Season 1: Ac: 59.5; Pl: 122.4 (P = 0.05)||3 meds [0 to unlimited] AUC for calendar weeks 20–37 in Season 1: Ac: 17.2; Pl: 36.8 (P = 0.14, NS)|
|Corrigan et al., 2005 (UK), study years 2002 and 2003. Long-term study||154||77/75||71/71 year 1, 66/60 year 2||5 months pre-seasonal each year||9 [0–3]. NR||7 [0 to unlimited]. Season 1: NR|
|SLIT with natural allergen drops|
|Sabbah et al., 1994 (France), study year 1991||58||29/29||29/29||4.25||8 [0–24]. NR. Ac better (P < 0.05) for four ISSs in pair-wise comparisons at some time points||5 meds [0 to unlimited]. NR. Significant decrease for 3 meds over a 5-week period in the SLIT group|
|Feliziani et al., 1995 (Italy), study year 1993||34||18/16||18/16||2 weeks pre-seasonal + 3 months co-seasonal||9 [0–27]. NR. AUC over 1 month: Ac: 109.7; Pl: 209.4 (P < 0.008)||3 [0 to NR]. NR. AUC over 1 month. Ac = 24.05; Pl: 76 (P = 0.002)|
|Clavel et al., 1998 (France), study year 1994||136||62/58||NR||7||6 [0–18]. NR. Median ± 95%CI AUC for rhinitis: Ac: 131 ± 22.71; Pl: 116.5 ± 36.68 (P = NS), for conjunctivitis: Ac: 53 ± 26.6 Pl: 43 ± 21.75 (P = NS)||3 [0 to NR]. NR. Median ± 95%CI, AUC over 13 weeks: Ac: 30 ± 22.96; Pl: 91.5 ± 31.58 (P < 0.01)|
|Hordijk et al., 1998 (The Netherlands), study year 1995||66||27/30||24/28||10||21 [0–63]. NR. Mean for pollen peak: Ac: 3.21 ± 3.05; Pl: 5.13 ± 3.60, (P = 0.03)||4 [0 to NR] NR. May: Ac: 22; Pl: 30; P = NS July: Ac: 29; Pl: 29; P = NS August: Ac: 18; Pl: 11; P = NS Sept: Ac: 11; Pl: 0; P = NS|
|Pradalier et al. 1999 (France), study year 1996||126||62/61||60/59||5||7 [0–21]. NR. Mean rhinitis SS: Ac:2.33 ± 1.61; Pl: 2.65 ± 1.97 (P = NS)||4 [0 to unlimited]. NR. AUC, Ac: 1.77 ± 2.27; Pl: 2.13 ± 2.74 (P = NS)|
|de Blay et al., 2007 (France), study year 2002||118||58/57||52/49||10||10 [0–30]. Mean AUC for rhinitis + conjunctivitis: Ac: 30.05 (sum of 22.26 ± 16.55 and 7.79 ± 9.28); Pl: 34.3 (sum of 23.12 ± 17.50 and 11.18 ± 10.82) (P = NS)||7 [0 to unlimited]. NR. AUC: Ac: 7.18 ± 11.59; Pl: 9.15 ± 10.79 (P = 0.11). CS: Ac: 39.31 ± 32.30; Pl: 45.31 ± 33.98 (P = 0.22)|
|Mosges et al., 2007 (Germany), study year 1998||102||48/53||42/48||9||7 [0–21]. NR. Median AUC: Ac: 180; Pl: 277 (P = 0.038)||5 [0 to unlimited]. NR. Median AUC: Ac: 13; Pl: 26 (P = 0.64)|
|Torres Lima et al., 2002 (UK), study years 1998–1999||56||28/28||26/23||12–18 months||9 [0–27] NR. Median AUC for 9 weeks: Ac: 1694; Pl: 2349 (P = 0.48)||5 [0–7], NR. Median AUC for 9 weeks Ac: 1418; Pl: 2569 (P = 0.19)|
|Smith et al., 2004 (UK), study years 1999–2000. Long-term study||186||121/59 (year 1)||45 (Ac)/46 (PI)/49 (Ac then Pl) (year 2)||1 and 2 years||10 [0–30]. NR. Season 1: NS for all symp.||3 [scale NR]. NR. P = NS|
|Worm, 2006 (Germany, Republic of Macedonia, Poland), study years 2003–2005. Long-term study||185||49/55||41/48||18||10 [0–30]. NR.||4 [NR]. NR. Only CS AUC scores given. For 2003 (baseline, before treatment): Ac: 501; Pl: 430 (P = NS); for 2005, Ac: 197; Pl: 452 (P = 0.0005)|
|Ott et al., 2008 (Germany), study years (4) NR. Long-term study||213||99/46||58/33||3 months per year for 3 years||9 [0–27]. NR. Mean ± SD changes for Season 1 relative to the baseline season: Ac: 0.03 ± 13.84; Pl: 1.49 ± 4.57 (P = 0.0366)||7 meds. total 0 to unlimited. Mean ± SD changes for Season 1 relative to the baseline season: Ac: 0.86 ± 11.72; Pl: 0.39 ± 3.04 (P = 0.3512)|
|Yuksel et al., 1999 (Turkey), study years 1996 and 1997. Paediatric. Long-term study||39||21/18||NR||15||8 [0–24]. NR, AUC lower for SLIT (P < 0.05)||3 [scale: number of medication days]. NR. Significant decrease in antihistamine consumption for Ac (P < 0.005). Nonsignificant for β-agonists|
|Wuthrich et al., 2003 (Switzerland), study year 2000. Paediatric. Long-term study||28||14/14||10/12||20||5 [0–15]. NR. AUC Ac: 105; Pl: 115 (P = NS).||3 [0 to unlimited]. Season 1 NR. AUC Season 2: Ac: 25; Pl: 75 (P = 0.05)|
|Bufe et al., 2004 (Germany), study years 1998–2001 but only 1999 DB. Paediatric. Long-term study||161||83/78||68/64 (after 1 year)||36||12 [0–36]. NR.||10 [0–100]. NR. After the first blinded year, no significant improvement in CS with SLIT|
|Rolinck-Werninghaus et al., 2004 (Germany), study years 1999–2001. Paediatric. Long-term study||97||49/48||47/45 after 1 year; 38/37 after 3 years||32||12 [each 0–3]. NR. AUC for nasal and ocular scores for 6-week pollen peak (adjusted for pollen count) in Season 1: Ac: 29.61; Pl: 20.84 (P = NS)||7 [0 to unlimited]. Mean for 6-week peak (adjusted for pollen count) in Season 1: Ac: 6; Pl: 3.18 (P = NR)|
|Roder et al., 2007 (The Netherlands), study years 2002–2004. Paediatric. Long-term study||204||91/77||64/50||24||5 [0–15]. NR. Pooled mean for 2002–2004. Ac: 3.1; Pl: 3.4 (P = 0.398)||7 [scale NR] NR. Pooled mean medication-free days results for 2002–2004: Ac: 69.3; Pl: 74.2 (P = NS)|
|SLIT with allergoid tablets|
|Palma-Carlos et al., 2006 (Portugal), study years 2001–2003. Long-term study||33||17/16||13/7||NR||3 [0–9]. NR. AUC Season 1: Ac: 196; Pl: 193 (P = NR)||6 Meds. NR but MS for Ac significantly lower (P < 0.02) in both seasons|
|Caffarelli et al., 2000 (Italy), study year NR. Paediatric||48||24/24||24/20||3||12 [0–36]. NR. Mean weekly SS: Ac: 9.5 ± 7.2; Pl: 14.5 ± 8.3 (P = 0.05)||4 [0 to unlimited]. NR. Mean weekly MS: Ac: 8; Pl: 9 (P = NS)|
|SLIT with natural allergen tablets|
|Durham et al., 2006 (Belgium, Denmark, Germany Sweden, Austria, Norway, the UK, Canada), study years 2002 (baseline/power calculation) and 2003||855||136 + 139 + 294/286||122 + 125 + 251/250||4.25||10 [0–30], Ac (75 000 SQ-T, >8 weeks pre-seasonal treatment): 2.513 ± 0.155; Pl: 3.176 ± 0.159, P = 0.002. NS for the other 2 doses||Meds and scale NR. Ac (75 000 SQ-T, >8 weeks pre-seasonal treatment): 1.612 ± 0.184; Pl: 2.260 ± 0.189, P = 0.012; NS for the other 2 doses|
|Dahl et al., 2006 (Austria, Denmark, Germany, Italy, the Netherlands, Spain, Sweden, the UK), study year 2005||634||316/318||274/272||>6 (16–35 weeks pre-seasonal + 2–12 weeks co-seasonal)||6 [0–18]. Ac: 2.36 ± 1.6; Pl: 3.37 ± 2.2, P < 0.0001||3 [0 to unlimited]. Ac: 1.5 ± 1.9; Pl: 2.4 ± 2.5, P < 0.0001|
|Didier et al., 2007 (Austria, Bulgaria, Czech Republic, Denmark, France, Germany, Hungary, Italy, Slovakia, Spain), study year 2005||628||157 + 155 + 160/156||139 + 133 + 141/146||6 (>4 months pre-seasonal and 1–2 months co-seasonal)||6 [0–18]: Ac. (500 IR) = 3.74 ± 3.142 (P = 0.0006 vs Pl), Ac. (300 IR) = 3.58 ± 2.976 (P = 0.0001), Ac. (100 IR) = 4.72 ± 3.141 (NS). Pl: 4.93 ± 3.229||2 meds. Scale NR. NR. Median (P25–P75) percent days with med, Pl: 19.72 (0.00–46.67)Ac (100 IR) = 15.19 (0.00–42.75) (p vs Pl: NR); Ac (300 IR) = 10.62 (0.0–29.77) (0.0194 vs Pl, Ac. (500 IR) = 10.53 (0.0–40.63) (NS)|
|Dahl et al., 2008 (Austria, Denmark, Germany, Italy, the Netherlands, Sweden, the UK), study years 2005–2006 (2005 published as Dahl et al., 2006). Long-term study||351||172/144||169/137||22||6 [0–18]. Ac: 2.4; Pl: 3.76, difference 1.36, 95%CI [0.86–1.86], P < 0.0001||4 [0 to unlimited]. Ac: 1.74; Pl: 3.19, difference 1.45, 95%CI [0.75–2.16], P < 0.0001|
|Wahn et al., 2009(France, Spain, Germany, Poland, Denmark), study year 2007. Paediatric||278||131/135||112/115||6||6 [0–18]. Ac: 3.25 ± 2.860; Pl: 4.51 ± 2.931, P = 0.001||3 [0 to NR]. Ac: 0.60 + 0.611; Pl: 0.79 + 0.647 (P = 0.0064)|
|Bufe et al., 2009(Germany), study year 2007. Paediatric||253||117/121||114/120||6.5||10 [0–30]. Ac: 2.67 ± 2.38; Pl: 3.17 ± 2.14 (P = 0.0195)||6 [0–66]. Ac: 2.13 ± 3.48; Pl: 2.53 ± 3.03 (P = 0.0156)|
|Authors, year of publication (countries), study year||Safety data*||Medication-driven SAEs?†||Compliance assessed?|
|SCIT with natural allergens|
|Varney et al., 1991 (UK), study year 1989||No precise data on TEAEs or LAEs. SystAEs: 10/13%||One in active group (shortness of breath, responded to intramuscular adrenalin)||NR|
|Zenner et al., 1997 (Germany), study year 1993||LAEs in 9.7% of active injections (not patients) and 2.1% of Pl injections. SystAEs: 20/12%||No||NR|
|Leynadier et al., 2001 (France), study year 1998||LAEs: 38/0%; SystAEs: 44/17%||No||NR|
|Frew et al., 2006 (UK), study year 2002||Early LAEs: 9% for pooled groups. Delayed LAEs: 42% for pooled groups. Early SystAEs: 32.5% high dose/21.2% low dose/16.5% Pl. Delayed SystAEs in 56% subjects (67%, 51%, 40%)||Nine (4.4%) subjects on 100 000 SQ-U had non-life-threatening grade 3 reactions (urticaria or asthma). No grade 4 reactions||NR|
|Durham et al. 1999, (UK), study years 1993–1995 but treatment 19891995 for some patients. Long-term study||NR||NR||NR|
|Dolz et al., 1996 (Spain), study years 1990–1992. Long-term study||Local reactions in four ‘patients’ and systemic reactions in seven Ac. patients. No reactions in Pl. group||Seven reactions requiring subcutaneous adrenaline||NR|
|Walker et al., 2001 (UK), study years 1996–1998. Long-term study||Mild delayed SystAEs during induction: Ac: 4 events; Pl: 5 events and during maintenance: Ac: 3 events and Pl: 0 events||No||NR|
|SCIT with allergoids|
|Drachenberg et al., 2001 (Germany and Austria), study year 1999||LAEs: 80/33%; SystAEs 17/12%||No||NR|
|Brewczynski & Kroon, 1999 (Poland), study years 1991–1993. Long-term study||Data on total numbers of incidents per injection, not % of patients affected. For season 1, LAEs: 114 out of 150 Ac, 30 out of 150 in Pl. SystAEs: 8 out of 150 Ac, 14 out of 150 in Pl.||No||NR|
|Corrigan et al., 2005 (UK). Long-term study||LAEs: 19.5% of actively treated patients; SystAEs:7/3%||No||NR|
|SLIT with natural allergen drops|
|Sabbah et al., 1994 (France), study year 1991||TEAEs 34/24%; LAEs and SystAEs NR||No||NR|
|Feliziani et al., 1995 (Italy), study year 1993||NR||No||NR|
|Clavel et al., 1998 (France), study year 1994||TEAEs NR; LAEs: 29/17%; SystAEs: none||No||NR|
|Hordijk et al., 1998 (The Netherlands), study year 1995||TEAEs not given; LAEs: 20 Events/25 Events; SystAEs: none||No||NR|
|Pradalier et al., 1999 (France), study year 1996||Patients with at least one AE for Ac/Pl : 27/19% (LAEs and SystAEs NR)||No||NR|
|de Blay et al., 2007 (France), study year 2002||LAEs, 69/40%; SystAEs: none||No||99.6 ± 3.8%|
|Mosges et al., 2007 (Germany), study year 1998||All local events, oral itching and burning more frequent in the SLIT group (P < 0.001) but similar in rhinitis and conjunctivitis||No||NR|
|Torres Lima et al., 2002 (UK), study years 1998–1999. Long-term study||TEAEs and LAEs NR; SystAEs: none||No||NR|
|Smith et al., 2004 (UK), study years 1999–2000. Long-term study||Patients with at least one AE (minor) Ac/Pl: 70/44%; fewer AEs in year 2 35/15% (no data on LAEs, SystAEs)||One episode of grade 3 generalized urticaria (Pl)||NR|
|Worm, 2006 (Germany, Republic of Macedonia, Poland), study years 2003–2005. Long-term study||TEAEs 73/38%; LAEs, 73/38%; SystAEs: none||No||NR|
|Ott et al., 2008 (Germany), study years (4) NR. Long-term study||TEAEs; 69/62%; no separate data for % LAEs and % SystAEs:||No||88%|
|Yuksel et al., 1999 (Turkey), study years 1996 and 1997. Paediatric. Long-term study||NR||No||NR|
|Wuthrich et al., 2003 (Switzerland), study year 2000. Paediatric. Long-term study||No AEs||No||NR|
|Bufe et al., 2004 (Germany), study years 1998–2001 but only 1999 DB. Paediatric. Long-term study||NR||No||Exclusions for poor compliance: Ac: 7; Pl: 7|
|Rolinck-Werninghaus et al., 2004 (Germany), study years 1999–2001. Paediatric. Long-term study||TEAEs Ac/Pl 49/27%; SystAEs: none||One (bronchial asthma) in active group||NR|
|Roder et al., 2007 (The Netherlands), study years 2002–2004. Paediatric. Long-term study||LAEs 39/17%; SystAEs: none||No||80/71%|
|SLIT with allergoid tablets|
|Palma-Carlos et al., 2006 (Portugal), study years 2001–2003||No SystAEs experienced by any study participant. LAEs: 12/0%||No||NR|
|Caffarelli et al., 2000 (Italy), study year NR. Paediatric||No AEs experienced by any study participant||No||NR|
|SLIT with natural allergen tablets|
|Durham et al., 2006 (Belgium, Denmark, Germany Sweden, Austria, Norway, the UK, Canada), study year 2003||TEAEs for Ac/Pl: 92% for 75 000/74%. LAEs: not reported SystAEs: not reported||One (uvula edema in the 25 000 SQ-T group)||NR|
|Dahl et al., 2006 (Austria, Denmark, Germany, Italy, the Netherlands, Spain, Sweden, the UK), study year 2005||TEAEs for Ac/Pl: 84/64%. LAEs: 88/22%; SystAEs: 22/1%||No||NR|
|Didier et al., 2007 (Austria, Bulgaria, Czech Republic, Denmark, France, Germany, Hungary, Italy, Slovakia, Spain), study year 2005||TEAEs for 300 IR /P 69/49%; SystAEs 50/14%; LAEs 19/13%||3 (none related to study medication)||Ac 300 IR: 87.7%; Pl: 95.5%, similar in other groups|
|Dahl et al., 2008 (Austria, Denmark, Germany, Italy, the Netherlands, Sweden, the UK), study years 2005–2006 (but 2005 published as Dahl 2006)||TEAEs for Ac/Pl: 51/41%. LAEs & SystAEs: NR||No||NR|
|Wahn et al., 2009 (France, Spain, Germany, Poland, Denmark), study year 2007. Paediatric||TEAEs for Ac/Pl: 85/82%. LAEs: NR SystAEs: NR||No||Ac: 94%; Pl: 95%|
|Bufe et al., 2009 (Germany), study year 2007. Paediatric||TEAEs for Ac/Pl: 87/83%, LAEs: NR. SystAEs:||No||Overall: 93%|
SCIT with natural grass pollen extracts
Study quality and treatment efficacy
Despite heterogeneity in the symptom scores (means, medians, area under the curve, challenge responses, etc.) and high variance, five of the seven trials showed significant efficacy for the active treatment in the first treatment season. The only large-scale, dose–response investigation was performed in a year-long trial on 410 randomized participants with two active dose levels (calculated cumulative dose: 207.6 and 20.76 μg, respectively) and a placebo group (20). There were dose-dependent reductions in the symptom score and (for the high dose only) the medication score (P = 0.0007). The remaining trials featured low to middling population sizes (from 28 to 87 randomized subjects). The SCIT modality was the first to be investigated in terms of long-term efficacy, with up to 84 months of treatment in a trial comparing 3 years of treatment (discontinuation) with 6 years (i.e. continuation for an additional 3 years) in 47 randomized patients (22). Dolz et al.’s (21) three-season study on 28 randomized patients reported significant (P < 0.001) active vs placebo differences for ocular symptoms in seasons 1, 2 and 3, for nasal symptoms in seasons 2 and 3 and for lung symptoms in seasons 1 and 2. Medication score results were similarly heterogeneous. In Walker et al.’s (23) two-season study, a significant active vs placebo difference was seen in the second season (P = 0.01).
All studies reported a higher proportion of adverse events (AEs) in SIT groups than in placebo groups. Systemic AEs requiring administration of subcutaneous adrenaline were observed (17, 21).
Varney et al. and Durham et al. (17, 21) asked the participants ‘How has your hay fever been during the past week?’ with a significant difference in favour of SIT. Frew et al. (20) reported a dose–response relationship for the RQLQ, with scores of 1.31 ± 0.14 for the high-dose group (P < 0.0001), 1.75 ± 0.17 for the low-dose group and 2.19 ± 0.17 for the placebo group. Walker et al. (23) reported that QoL was significantly less impaired in the SIT group [median difference (95% CI): 0.8 (0.2, 1.5) (P < 0.01)].
SCIT with modified grass pollen allergens (allergoids)
Study quality and treatment efficacy
Population sizes were small to moderate: 81 SIT-treated patients for Drachenberg et al., 10 for Brewczynski & Kroon and 77 for Corrigan et al. (24–26). Drachenberg et al. (24) reported a significant (P < 0.05) active vs placebo difference for the total symptom score after 1 month of treatment but no significant difference for medication scores. In the first season of their 3-year study, Brewczynski & Kroon (25) observed a (borderline) significant (P = 0.05) active vs placebo difference for the symptom score but not for the medication score. Although Corrigan et al.’s (26) two-season study failed to report separate symptom and medication scores for season 1, there was a statistically significant difference in the pooled score for two seasons.
Brewczynski & Kroon (25) reported that in season 1, local AEs occurred after 114 of the 150 injections in the SIT group, vs 30 of 150 in the placebo group. For systemic AEs, the values were 8 and 14 injections in the active and placebo groups, respectively. Drachenberg et al. (24) reported more local AEs than systemic ones and more events in the active group than in the placebo group (80/33% and 17/12%, respectively). Corrigan et al. (26) described a similar picture, with delayed systemic reactions in five SIT-treated patients and two placebo-treated patients.
Corrigan et al. (26) observed a greater RQLQ improvement in the active group than in the placebo group (−0.74 and −0.78 points, respectively, P = 0.0252). Brewczynski & Kroon (25) asked general questions on patient satisfaction and symptoms severity but no firm conclusions could be drawn. Drachenberg et al. (24) did not report on patient QoL or satisfaction.
SLIT grass pollen natural extracts (drops)
Study quality and treatment efficacy
We identified 16 trials (from 1994 to 2008). Population sizes ranged from 18 to 121 SIT-treated subjects (28, 35). In the seven studies in adults, the treatment duration ranged from 2 weeks of preseasonal administration and 3 months of co-seasonal administration (28) to essentially year-long treatment (30). The maintenance doses of major allergen varied greatly from one study to another, from 0.5 μg Phl p5 per day to 40 μg per day (36, 39, 41). Likewise, cumulative doses in single-season studies ranged from 294 to 3390 μg (29, 30). These variations may have contributed to the heterogeneity of the trial outcomes, with variously (i) a significant difference in favour of SIT in the median total symptom score over the whole pollen season (33), (ii) an effect on medication scores (29), (iii) an effect on symptom scores for the pollen peak or at certain time points (27, 28), (iv) an improvement in investigator-assessed allergic complaints only (30) or (v) no SIT effect on efficacy parameters (31). The role of asthma status was mentioned by de Blay et al. (32) in a population of 118 randomized subjects; although no significant, overall effects of SIT on efficacy scores were apparent, a subgroup analysis revealed a significant (P < 0.045) symptom reduction in nonasthmatic patients.
Four adult studies addressed the long-term and/or season-to-season effects of SLIT drops and are described further below. All five paediatric SLIT drops trials were multi-season, with treatment durations ranging from 15 months to 3 years (38, 40).
The most frequently described AEs in children and adults were local oral symptoms, followed by abdominal pain, nausea and diarrhoea. These events were usually mild to moderate and did not require dose adjustment. Systemic AEs (asthma, rhinitis, urticaria and angio-oedema) were rarely observed, and no life-threatening AEs or fatalities were described.
Only Röder et al. (42) scored paediatric and adolescent versions of the RQLQ but did not find any significant active vs placebo differences.
SLIT allergoid tablets
Study quality and treatment efficacy
Only one adult study and one paediatric study were selected; both featured low sample sizes (17 and 24 SIT-treated subjects, respectively) (43, 44). In a two-season study, Palma-Carlos et al. (43) reported a significant season-to-season decrease in the symptom score for the active group only. In children, Caffarelli et al. (44) reported a borderline significant (P = 0.05) decrease in the symptom score and no difference in the medication score.
Few or no local AEs were reported (12% in the active group in Palma-Carlos et al.), and no systemic AEs were observed (43).
Neither study provided feedback on QoL or patient satisfaction.
SLIT grass pollen natural extracts (tablets)
Study quality and treatment efficacy
The five identified studies involved a total of over 3600 randomized subjects (SIT and placebo) and 1605 SIT-treated subjects – more than for all the other selected SIT trials. All the trials were ‘large’, as defined by Durham (61). Two studies (Durham et al. and Didier et al.) featured three different allergen doses and showed a dose–response relationship (45, 47). Durham et al. (45) enrolled a total of 855 participants in 55 centres to test three doses (2500, 25 000 and 75 000 SQ-T) of once-daily GRAZAX®Phleum pratense pollen tablets (ALK-Abello, Hørsholm, Denmark) for a mean duration of 18 weeks. The 75 000 SQ-T group showed significant improvements in the efficacy scores during the pollen peak and in patients with at least 8 weeks of preseasonal treatment. Subsequently, Dahl et al. (46, 48) included 634 patients in a long-term study of the daily 75 000 SQ-T tablet with three treatment seasons (starting in 2005) and two follow-up seasons. For the first season, a highly significant difference (P < 0.0001) in favour of SIT was reported for the mean rhinoconjunctivitis score, the mean medication score and the proportion of ‘well’ days. Significant improvements in favour of SIT were also found for all individual symptom scores (ISSs), the number of symptom- and medication-free days and the medication score.
With 628 randomized patients, Didier et al. studied three doses (100, 300 and 500 index of reactivity (IR)) of an ORALAIR® 5-grass pollen tablet (Stallergenes, Antony, France). The 500 and 300 IR doses (but not the 100 IR dose) significantly and similarly decreased the mean difference in the rhinoconjunctivitis total symptom score (RTSS) vs placebo (P = 0.0006, 0.0001 and 0.46, respectively), the mean and median medication score, the median number of medication-free days and all ISSs, suggesting a dose-dependent/plateau effect (47).
The two paediatric trials featured large population sizes (278 and 253 randomized subjects, respectively) (49, 50). Wahn et al. (49) studied 4 and 2 months of preseasonal and co-seasonal treatment with an ORALAIR® 300 IR 5-grass pollen tablet. The season’s mean ± SD RTSS was lower in the SIT group (3.25 ± 2.860) than in the placebo group (4.51 ± 2.931). Bufe et al. (50) studied a GRAZAX® 75 000 SQ-T Phleum pratense pollen tablet group whose mean RTSS (2.67 ± 2.38) was significantly lower than that of the placebo group (3.17 ± 2.14, P = 0.0195). All trials also found significant differences in favour of SLIT for one or more ISSs, the rescue-medication score and/or days with/without rescue medication.
All seven studies reported on safety in detail; the principal AEs were mild, local and transient and none required adrenaline administration. Treatment-related SAEs were not observed. The safety profile in the paediatric tablet studies was similar to that seen in adult studies.
Durham et al. (45) reported a significant improvement (20%vs placebo) in the RQLQ scores with the 75 000 SQ-T dose. Dahl et al. observed a significantly greater percentage of improved patients in the SIT group (82%vs 55% in placebo, P < 0.0001) (46) – a result that was confirmed by patient assessment and RQLQ scores (33% better for SIT, P < 0.0001) in the second year (48). Didier et al. (47) recorded a favourable patient evaluation for SIT and a 20% improvement in the RQLQ score (P < 0.0031) for the pooled 300 and 500 IR groups.
We identified a total of 15 ‘long-term’ studies: three studies for SCIT with natural allergens, two for SCIT with allergoids, nine for SLIT drops and one with SLIT natural extract tablets but none for SLIT with allergoid tablets.
All three long-term SCIT natural extract trials suggested that the effect of SIT was sustained during treatment and after treatment cessation (21–23). In Brewczynski & Kroon’s (25) 3-year study of allergoid SCIT, all subjects received SIT in the second and third years and so firm conclusions on long-term efficacy could not be drawn. Corrigan et al. (26) reported that allergoid SCIT produced greater combined symptom and medication score reductions (vs placebo) in the second season (48.2%) than in the first (26.6%).
In SLIT drop studies, Worm et al. (36) treated subjects throughout two pollen seasons. After equivalent scores in the baseline 2003 season; there was a significant combined score reduction in 2005 in the ITT and PP SIT populations (P = 0.0023 and 0.0005, respectively). Significant benefits were identified for nose running (P < 0.001) and sneezing (P < 0.05) scores in the second season. The study by Ott et al. (37) involved a baseline season, co-seasonal treatment for three consecutive pollen seasons and a carry-over season. A significant active vs placebo difference in the symptom score was seen in all three treatment seasons and in the carry-over season (P = 0.0366, 0.0235, 0.0004 and 0.0153, respectively), whereas the difference in medication intake was not significant. The two-season study by Smith et al. (35) featured an observation season (1998) and a partial cross-over with three treatment groups: 2 years of active treatment, 2 years of placebo or a year of active treatment and a year of placebo. The total symptom scores in the first analysed season was not tested but none of the ISSs showed a significant improvement.
All the selected paediatric SLIT drop trials were long-term but only one reported a significant effect of SIT on the symptom score in 39 randomized subjects from March 1996 to August 1997 (38). Wüthrich et al. (39) investigated 28 randomized subjects in 1999 and 2000 and reported a significant SLIT vs placebo-medication score difference in the second year but did not comment on the symptom score. Only, the first year of Bufe et al.’s (40) 3-year study was double-blind but there were no significant SLIT vs placebo differences. The 3-year trial by Rolinck-Werninghaus et al. (41) was fully double-blind but did not report season 2 scores or the results of direct SLIT vs placebo comparisons for each season. Röder et al.’s 2007 (42) medium-sized, 2-year trial with a primary care recruitment mode did not observe any significant SIT vs placebo efficacy differences.
The results of the second year of the SLIT tablet study initiated by Dahl et al. (46, 48) confirmed the year 1 findings, with maintenance of a significantly lower RTSS in the SIT group vs placebo [2.36 ± 1.6 vs 3.37 ± 2.2, respectively (P < 0.0001)].
The goal of this article was to use an EBM approach to determine the level of evidence for each commercially available SIT modality for the treatment of grass pollen SAR. Evidence-based medicine starts when a patient and a physician have to take a healthcare decision together and continues when the physician admits to the presence of gaps in his/her knowledge. We sought to help the physician to fill that gap by evaluating the validity and relevance of the published evidence.
Even with our strict focus on DBPC RCTs with a single type of allergen (grass pollen), we observed high variability in patient numbers, scoring systems, dosing regimens and treatment durations. In addition to the above-mentioned meta-analyses (the results of which we are not seeking to challenge), another way of investigating this variability would consist of an individual patient data (IPD) meta-analysis with adjustment for confounders at the individual level and the study level. However, obtaining IPD would probably only feasible for a subset of (mainly recent) studies and would create study selection bias. Furthermore, a standardized response criterion would be very difficult to define. In the absence of IPD, key questions can nevertheless be posed by adopting an evidence-based approach. How many patients were randomized? Were the active and placebo groups equivalent on randomization? What were the ITT and PP populations? How many study countries and centres were involved? Were clinical efficacy criteria scored throughout the whole pollen season? Were medication scores and symptom scores reported separately? How homogeneous were the results? How detailed was the safety reporting? We have attempted to summarize the overall results of our review in table form (Table 3).
|SIT modality||Total pooled ITT SIT-treated patients (number of studies)||‘Large’ studies*||International studies†||Adult studies with a significant effect of SIT‡||Paediatric studies with a significant effect of SIT‡||Studies evaluating patient opinion and/or QoL|
|SCIT – allergoids||168 (in 3 studies)||0 of 3|
|1 of 3|
|2 of 3|
|No paediatric studies||1 of 3|
|SCIT – natural extracts||440 (in 7 studies)||1 of 7|
|0 of 7|
|5 of 7|
|No paediatric studies||4 of 7|
|SLIT drops – natural extracts||906 (in 16 studies)||0 of 16|
|1 of 16|
|4 of 11|
|1 of 5|
|3 of 16|
|SLIT tablets – allergoids||41 (in 2 studies)||0 of 2|
|0 of 2|
|1 of 1|
|0 of 1|
|0 of 2|
|SLIT tablets – natural extracts||1605 (in 5 studies)||5 of 5|
|4 of 5|
|3 of 3|
|2 of 2|
|3 of 5|
Despite the long history of SCIT, only seven studies met our selection criteria. Furthermore, only one trial stood out in terms of its large sample size, dose ranging design, detailed reporting, investigation of patient QoL and overall satisfaction (20). Most SCIT natural extract trials reported good efficacy on symptom scores but safety remains an issue. Our selection criteria did not identify any SCIT trials with grass pollen natural antigen extracts or allergoids in paediatric populations.
Considering that there are few commercially available SIT products based on allergoids (whether for subcutaneous injection or as sublingual formulations), only five studies met our selection criteria (three in SCIT and two with SLIT tablets). The low cumulative number of subjects in these studies means that conclusions on the efficacy of allergoid preparations are limited by the amount of published data. Modern, large-scale, DBPC RCTs evaluating the efficacy of allergoids could increase the level of evidence for these preparations in SIT. The SCIT allergoid studies reported some systemic SAEs and gave little information on patient QoL and values, thus restricting the application of an EBM approach.
The SLIT drop modality featured the highest number of trials (16, including five in children) and the second highest total number of patients, although individual trial sample sizes, doses, regimens and treatment effects were very variable. Only, one of the five paediatric studies reported a significant effect of SIT on symptoms (38); this may be related to the relatively low daily doses evaluated (from 0.5 to 21 μg), when compared with those employed in SLIT drop or tablet studies in adults. In addition, safety was high and well documented, with mainly minor, local reactions and no treatment-driven, systemic SAEs.
Registered SLIT tablet formulations for treating grass pollen SAR have recently been investigated in large, methodologically rigorous trials with strong research evidence. Patient values and QoL have been addressed. Safety data were good: AEs are mainly local and resolve rapidly. Today’s increasingly demanding regulatory requirements will oblige SIT producers to move towards trials with sufficient statistical power and the extensive collection of safety and patient values data. Until recently, the level of evidence for SLIT in paediatric populations was low. After the positive findings of a 2006 meta-analysis (10), a 2008 review by Röder et al. (14) stated that ‘there is at present insufficient evidence that immunotherapy in any administration form has a positive effect on symptoms and/or medication use in children and adolescents with allergic rhinoconjunctivitis’. Since the publication of Röder et al.’s review, efficacy has been unambiguously demonstrated in the two large, paediatric Phase III SLIT tablet studies (49, 50). Overall, the five tablet studies demonstrate that SLIT is becoming a pharmaceutically robust, evidence-based treatment for grass pollen allergy.
Understandably, our approach suffers from limitations, including those common to all systematic reviews (publication bias, irretrievable full texts and so on). Many studies performed in the 1980s and 1990s did not report extensively on design and outcomes. Stricter regulatory and ethical requirements have increased the level of evidence associated with the last-developed modality (natural extract tablets). Furthermore, all but one of the investigating centres were located in Europe and so presumably investigated primarily Caucasian populations.
Our approach illustrates the fact that there are many ways of evaluating clinical trials – on the basis of professional consensus, using national or international guidelines or with other methodological initiatives, such as the Grading of Recommendations Assessment, Development and Evaluation approach (62). Despite the undoubted value of meta-analyses of clinical trials with significant clinical and methodological heterogeneity, there is a need for rigorous, large-scale RCTs with enough statistical power to detect clinically relevant differences and which are designed to give precise answers to outstanding clinical questions. The advantageous combination of this rigorous approach with other methodological advances aimed at reducing heterogeneity within SIT studies (such as improved symptom scoring and patient selection criteria) should notably result in better definition of the true level of disease severity experienced by SIT and placebo-group populations.
In conclusion, we hope to have initiated a transition towards an EBM approach to SIT by taking account of important trial parameters and outcomes. Even though several aspects of SIT remain to be described in detail (optimal regimen, onset of action, long-term and carry-over efficacy, effects on polysensitization and asthma, etc.), our approach suggests that SLIT tablets for grass pollen SAR provide the specialist physician with the strongest levels of evidence for drawing reliable conclusions.