Results of the search

Twenty records were identified by the search of which four were not relevant and were excluded. Sixteen full text reports were assessed and one was excluded as it was not a randomised trial. Eight reports of five studies were eligible for inclusion in the review. Additionally, there were seven studies that still appeared to be ongoing: one (Smart 2008) was a communication about an ongoing trial and the results were not yet available in the public domain at the time of the search; one study (NCT00762047 2009) appeared to have been terminated in 2009 (reasons unknown, trialists contacted for details); and three were ongoing studies for EAS injuries using cell therapy. The flow of study selection is summarised in Figure 1.

Figure 1

---

---

PRISMA study flow diagram,

Included studies

The five randomised trials (Graf 2011; Maeda 2008; Siproudhis 2007; Tjandra 2004; Tjandra 2009) included a total of 382 patients. Further details are in the Characteristics of included studies table.

The participants in all five RCTs had internal anal sphincter (IAS) dysfunction or passive faecal incontinence and were reported to have failed previous conservative management with anti‐diarrhoeal drugs and pelvic floor muscle training or biofeedback (although this was not standardised and generally not reported in any detail).

The interventions tested were:

• hydrogel cross‐linked with polyacrylamide (Bulkamid) (Maeda 2008);

• porcine dermal collagen (Permacol) (Maeda 2008);

• polydimethylsiloxane elastomer implants (Siproudhis 2007);

• silicone biomaterial (PTQ) injected between the two anal sphincters (Tjandra 2009);

• carbon‐coated beads injected into the anal submucosa (Durasphere) (Tjandra 2009);

• two different injection techniques (endoanal ultrasound guidance or digital guidance) using a silicone biomaterial (PTQ) (Tjandra 2004);

• dextranomer in stabilised hyaluronic acid (NASHA Dx) (Graf 2011).

The control intervention in two trials were:

• saline (Siproudhis 2007);

• sham injection (mimicking active treatment but without injection of any substance) (Graf 2011).

The injection sites were:

• under the anal submucosa (Graf 2011; Tjandra 2009)

• in between the two anal sphincters (Maeda 2008; Siproudhis 2007; Tjandra 2004; Tjandra 2009).

Excluded studies

See the Characteristics of excluded studies table. One study was not an RCT (Dehli 2007).

Allocation

Two trials gave sufficient details about the randomisation process to suggest freedom from bias (sequence generation in a central registry) (Graf 2011; Tjandra 2009). The quality of allocation concealment was judged to be unclear in three trials.

Blinding

In two trials (Graf 2011; Siproudhis 2007), the patients and the outcome assessors were blinded to treatment allocation. The other three trials did not specify whether the assessor was different from the person involved in the treatment.

It was not possible to blind the surgeons in any of the trials as the interventions were different.

Incomplete outcome data

Selective reporting

Review authors considered there was no evidence of selective reporting.

Other potential sources of bias

There are six ongoing trials (Dehli 2007; Dehli 2009; Draganic 2008; EUCTR2010‐021463‐32‐AT 2011; Michot 2012; Norderval 2012). Five studies are ongoing with three of them aiming to regenerate muscles by cell therapy for external anal sphincter injuries, hence they are not under the scope of the current review. One study has been completed and is under preparation for publication (personal communication).

1. Injection of bulking agent versus placebo or sham injection

Two trials addressed this comparison.

Siproudhis et al (Siproudhis 2007) compared a silicone biomaterial (polydimethylsiloxane elastomer, PTQ) with a (control) normal saline injection into the intersphincteric space. The trial was too small to detect differences in any outcomes, such as failure rate (17/22, 77% versus 16/22, 72% in the control group, RR 1.06, 95% CI 0.75 to 1.50; Analysis 1.1); use of pads (Analysis 1.3) or Wexner score (Analysis 1.5). In addition, the subjective perception of treatment success was low in both groups (23% versus 27%, P = 0.73).

The study by Graf et al (Graf 2011) compared injection of dextranomer stabilised in hyaluronic acid (NASHA Dx) against sham injection (injection without actual substance). The primary endpoint was a response to treatment which was defined as a 50% or greater reduction in the number of incontinence episodes. There was a statistically significant difference between the treatment and control groups at six months (65/136, 48% versus 48/70, 69% participants not improved, defined as less than 50% reduction in incontinence episodes, RR 0.70, 95% CI 0.55 to 0.88; Analysis 1.2). Participants in the active group had statistically significantly more incontinence free days at six months (3.1 days compared with 1.7 in the sham treatment group, MD 1.40 days, 95% CI 0.33 to 2.47; Analysis 1.4). However, this difference was not shown at three months. Secondary endpoints such as number of faecal incontinence episodes, CCFIS at three and six months and FIQOL at 12 months did not show any difference except for one domain of FIQOL between the active and sham treatment groups. While there were fewer adverse events in the control group (128/136, 94% in the active group versus 29/70, 41% in the control group, RR 2.27, 95% CI 1.71 to 3.01; Analysis 1.6.2), only two in the active group were regarded as serious (rectal and prostate abscesses).

2. Injection of bulking agent versus conservative treatment

No trials were found.

3. Injection of bulking agent versus another minimally invasive or surgical intervention

No trials were found.

4. One type of bulking agent versus another type

Two trials addressed this comparison (Maeda 2008; Tjandra 2009).

In the trial by Tjandra and colleagues (Tjandra 2009), injection of the silicone biomaterial polydimethylsiloxane elastomer (PTQ™) and carbon‐coated beads (Durasphere®) were compared. While participants in both groups reported an improvement in Wexner's score and quality of life over the study period up to 12 months, the silicone material was better than the carbon‐coated beads in terms of: faecal incontinence at six and 12 months (for example at 12 months, RR 0.15, 95% CI 0.04 to 0.60; Analysis 2.1); Wexner's incontinence score at six and 12 months (for example at 12 months, MD ‐3.20, 95% CI ‐4.91 to ‐1.49; Analysis 2.2); and quality of life (lifestyle) index at six months (MD ‐0.56, 95% CI ‐1.01 to ‐.011; Analysis 2.3). There were more adverse effects (4/20 in the silicone group versus 8/20 in the carbon‐coated bead group; Analysis 2.4); one serious event occurred in the carbon‐coated bead group only (Analysis 2.5). It was assumed that there were no drop‐outs, though this was not explicitly stated.

Another trial with only 10 participants (Maeda 2008) compared injection of hydrogel cross‐linked with polyacrylamide (Bulkamid™) and porcine dermal collagen (Permacol™). In the hydrogel group incontinence scores improved from a median of 15 to 12 at six weeks and this was maintained at six months. In the porcine dermal collagen group, the baseline median score improved from 16 to 14 at six weeks but deteriorated to 15 at six months. The trial was too small to detect differences between the groups in terms of faecal incontinence becoming worse (Analysis 2.1), the overall St Mark's incontinence score, or the faecal incontinence quality of life score, but data were not provided in a form suitable for analysis (medians and no SDs).

5. One technique of administration of agent compared to another technique

One trial (Tjandra 2004) compared a silicone biomaterial, polydimethylsiloxane elastomer (PTQ™), injection under endoanal ultrasound guidance versus digital (finger) guidance. Using a criterion of greater than 50% decrease (improvement) in the Wexner scale, the ultrasound technique was better than the digital technique (RR 0.53, 95% CI 0.31 to 0.92; Analysis 3.1, Analysis 3.2) and improved the Quality of Life (lifestyle) index at six months (MD ‐0.60, 95% CI ‐0.92 to ‐0.28; Analysis 3.3). The only reported adverse effect, discomfort at the injection site, was rare (2/42 and 4/40 participants; Analysis 3.4). However, many of the data were reported in a form not suitable for analysis (medians and no SDs), the follow up was short (12 months), and there was a high drop‐out rate (by 12 months, 32/42 and 35/40 respectively).

6. Larger volumes of agent versus smaller volumes

No trials were found.

7. Safety data in included RCTs and other non‐randomised studies

Adverse effects were reported in four of the five included trials (Graf 2011; Siproudhis 2007; Tjandra 2004; Tjandra 2009). In the randomised trials included in this review, one trial (Siproudhis 2007) reported 18 adverse effects in nine people (Analysis 1.6). Six out of 22 people injected with silicone biomaterial (PTQ™) complained of pain at the implant site, whilst two out of 22 people in the saline group complained of pain. Anal inflammation was noted in two people in the silicone biomaterial injection group compared to none in the normal saline group. The severity of these adverse effects was categorised as 'not serious' but the duration of the events was not reported. In the trial reported by Graf et al (Graf 2011), patients treated with dextranomer stabilised in hyaluronic acid (NASHA Dx) reported 128 adverse events (number of patients not stated) whilst 29 events (number of patients not stated) were recorded in the sham treatment group (Analysis 1.6). The most common event was proctalgia (19/136 versus 2/70) followed by injection site bleeding (7/136 versus 12/70) and rectal haemorrhage (10/136 versus 1/70). Two serious events were reported in the active treatment group, one was a prostatic abscess resolved by antibiotic treatment and the other was a rectal abscess which required surgical drainage. The trial comparing silicone biomaterial (PTQ™) and carbon‐coated beads (Durasphere®) (Tjandra 2009) reported no septic or significant complications in patients injected with PTQ™, although there were four occurrences of bruising. In contrast, patients who had a Durasphere® injection complained of rectal pain (1/20), erosion of rectal mucosa (2/20) and type III hypersensitivity reaction in small joints and skin (1/20) (Analysis 2.4). One of these type III hypersensitivities was judged to be serious and resulted in hospital admission for seven days (Analysis 2.5). The trial comparing injection techniques (Tjandra 2004) reported minor discomfort in two of 42 patients who had injections with ultrasound guidance and four of 40 patients with digital guidance (Analysis 3.4); in one patient this persisted for six weeks. There were no reports of other serious adverse effects (Analysis 3.5).

The occurrence of complications was also investigated in all non‐randomised studies found to date and is summarised in Table 1. In general, the complications were poorly reported.The severity and duration of complications were rarely reported and in some studies there was no mention of whether or not there had been any complications. The complications were mostly minor (discomfort, pain, bleeding, leakage of injected material). One rectovaginal fistula was reported although it was not clear if this was related to the injection. The reason for discontinuation of one study (NCT00762047 2009) is unclear and there is no information on adverse events in this study.

Most studies do not appear to have systematically and prospectively collected data on adverse events, but instead reported those which came to the investigators' attention.