An 8‐year, single‐centre experience of primary image‐guided insertion of ‘button’ gastrostomy catheters: Technical and clinical results

‘Button’ gastrostomy insertion is traditionally a two‐step procedure with an initial longer gastrostomy tube inserted followed by placement of the shorter ‘button’ gastrostomy in 6 weeks when the track is mature. The aim of this study is to assess whether the placement of a Button gastrostomy de novo is a safe and effective method of radiologically inserted gastrostomy (RIG) insertion.


Introduction
Gastrostomy insertion is an important alternative for providing enteral nutrition to patients who cannot ingest food orally. Indications for gastrostomy insertion include head and neck tumours (both in the pre-and postoperative setting), neurological disorders (motor neuron disease, stroke patients, neurosurgical patients and patients post neurological trauma), oesophageal tumours and other patients at high risk of aspiration.
Gastrostomy insertion can be performed surgically, via an endoscopic approach (percutaneous endoscopic gastrostomy (PEG)) or via a radiological approach (radiology inserted gastrostomy -RIG) using fluoroscopic guidance. We have a close relationship with surgeons/physicians and dietetic department and as our hospital is the national neurosurgical centre and also has a large head and neck oncology service, there is a significant volume of referrals for gastrostomy insertion. In our institute, RIG insertion is the preferred method of gastrostomy insertion for patients with neurology disorders (due to risk of aspiration with PEG) as well as for patients with large oesophageal and head or neck tumours (endoscopy often not possible).
The safety and efficacy of RIG insertion is well documented, [1][2][3][4] but the optimal technique used is still debated. A RIG 'button' feeding tube is traditionally a two-step procedure whereby a longer gastrostomy is sited during the initial intervention and a shorter 'button' gastrostomy is sited at a later date, usually at 6 weeks, once the tract has matured. In our institution, RIGs were performed with a 'button' at the time of the primary intervention, and thus negate the requirement for another procedure.
The aim of our study was to present our experience of this method of primary 'button' RIG insertion including the success rate, complications, exchange rate and 30day mortality.

Methods
Data collection was carried out over a total period of 8 years between 1 January 2012 and 1 January 2020. A follow-up period of at least 1 year was observed up to 1 January 2021. All patients who had a 'button' RIG attempt during this period were identified and reviewed using our Picture Archive and Communication System (PACs) and electronic patient charts.
Technical success rate, complications, exchange rate and 30-day mortality were recorded.
Complications were recorded as either major or minor. No consensus exists as to the precise nomenclature of complications following gastrostomy. Major complications were recorded as mortality, tube malposition within the peritoneal cavity, colon injury or a major bleed (which required intervention or a blood transfusion). Minor complications were considered when the complication did not result in significant therapy or intervention and required only conservative management or observation. Examples of minor complications included minor bleeding, skin irritation or minor site infection.
In addition, 30-day mortality was recorded including cause of death and if mortality was directly related to the RIG insertion.
Institutional review board approval was not required for this type of study and the principles of the Declaration of Helsinki were followed.

Technique
The technique for gastrostomy insertion has been previously described in the literature. 2 Routine preoperative barium administration is no longer routine practice and, like most interventionalist, gas outlining the transverse colon is felt to be sufficient to avoid the colon. A nasogastric tube is inserted prior to the procedure and this is used to insufflate the stomach. Patients are visited on the ward prior to the procedure to ensure suitability and to obtain informed consent. Patients are fasted overnight. Local anaesthetic and conscious sedation, usually a combination of intravenous Fentanyl (Fentanyl, Mer-curyPharma, Dublin, Ireland) and Midazolam (Hypnovel, Cheplapharm Arzneimittel GmbH, Greifswald, Germany), are administered in incremental doses as required. Intravenous hyoscine butylbromide (Buscopan, Sanofi, Reading, United Kingdom) 20 mg is also administered to reduce smooth muscle contraction and peristalsis. Prophylactic antibiotics were not routinely administrated.
Fluoroscopic images are performed to identify the colon, and the stomach is then inflated (Figs 1,2). We have been using the MIC-KEY* (Fig. 3) gastrostomy feeding tube (Avanos Medical, Alpharetta, Georgia, USA) in all patients since 2012. Under fluoroscopic guidance 3-4 T-fasteners (gastropexy sutures which come in the gastrostomy kit) are inserted into the stomach in a triangular or square manner approximately 2 cm apart from each other. Positioning of the T-fasteners is confirmed on both AP and lateral projections (Fig. 4). Within the centre point of the T-fasteners, an incision is made and an 18-g vascular needle is used to enter the stomach. An Amplatz Super Stiff (Boston Scientific, Marlborough, Massachusetts) guidewire is then coiled in the stomach (Fig. 5). Over the guidewire, a telescoping dilator ( Fig. 6) with an integrated peel-away sheath (included in the kit) is inserted and dilated to the 'red' dilator, the telescoping dilator is removed and a graduated measuring device with a balloon (part of gastrostomy kit) inflated in the stomach. The red dilator is the size compatible with the measuring tube as per manufacturer; however, it is our experience that often one dilator size up is required to comfortably insert the measuring device. The measuring tube is then pulled back until it abuts the anterior wall of the stomach, a measurement is then taken to accurately measure the tract length. A button device 5 mm longer than the measured track length is chosen. The telescoping fascial dilator, with an integrated peel away sheath, is reinserted and the track fully dilated until peel-away sheath is in the stomach (Fig. 5). The track is dilated to a size 4 Fr larger than the 'button' to be inserted in order to accommodate the larger retention balloon at the end of the 'button' gastrostomy. A 14 Fr or 18 Fr button gastrostomy is then inserted through the peel away sheath and the balloon is inflated with 7-10 mL of sterile water. Button position within the stomach is confirmed with the injection of contrast material on both AP and lateral projections (Figs 7,8).
Post 'button' insertion, patients are kept nil by mouth and the RIG is not used for a 24-h period. The IR team see the patient on the ward the next morning, T-fasteners are cut and use of the 'button' gastrostomy is allowed if the patient is without pain and the abdominal examination is normal. Initially, 40 mL an hour, for 4 h, of sterile water is administered via the RIG and if the patient tolerates this and vital signs remain normal, the proper feeding regime is commenced.

Results
In total, 482 patients underwent primary 'button' radiologically inserted gastrostomy over the 8-year period.
The mean age of the patients was 59.55 years (range 18-88 years old) with 290 men (60.2%) and 192 women (39.8%). Indications for RIG insertion were broken into four categories: neurological (the majority of which were neurosurgical, stroke and motor neurone disease patients), head and neck malignancy, oesophageal malignancy and miscellaneous indications (Table 1).
Primary placement of a 'button' gastrostomy was successful in 97.1% (468/482) patients. Of the 14 patients in whom primary 'button' gastrostomy failed, 7 (50%) had pigtail gastrostomy tubes inserted but were    considered as primary failures as a 'button' gastrostomy was not inserted as planned. In the latter 7 /14 patients, standard gastrostomy tubes or pigtail catheters were placed because of: increased BMI resulting in a long tract length of >5 cm (n = 4) which negated 'button' gastrostomy placement as button length does not exceed 5 cm; well-developed abdominal musculature (n = 1) which    prevented track dilation to the appropriate diameter for button insertion and required placement of a 12 Fr pigtail catheter followed by track dilation and 'button' insertion in the mature track 6 weeks later; high position of stomach above xiphisternum (n = 1), with no room for Tfasteners to be placed and therefore a 12 Fr pigtail catheter was placed and button tube placed after 6 weeks; a button tube was recognized to be outside the stomach on final check tubogram in one patient, this was recognized at the time of the procedure, the button tube was removed, a guide wire was manipulated back into the stomach and after appropriate track dilation, a 12 Fr pigtail catheter was placed with eventual button placement 6 weeks later. In a further seven patients, no gastrostomy tube could not be placed because of: the position of the colon which overlay the stomach (n = 3), a VP shunt which passed in front of the stomach (n = 1), previous hernia mesh repair overlying the stomach (n = 2) and extreme patient agitation (n = 1).
A major complication was recorded in 0.8% (n = 4) of cases. Major complications included bleeding that required intervention (endovascular embolization in 1 and blood transfusion without surgical or IR intervention in 1), malposition of tube in the peritoneal cavity with subsequent feeding (patient had laparotomy and peritoneal washout) (n = 1) and injury to colon (patient had a laparotomy and primary repair of transverse colon) (n = 1) ( Table 2). Minor complications were recorded in 1.7% (n = 8), four patients reported minor bleeding and four patients reported minor site infections/irritation.
A 30-day mortality of 1% (n = 5) was recorded, of which one was a direct complication of the 'button' RIG insertion. The latter patient died from peritonitis as a result of tube malposition within the peritoneal cavity despite laparotomy and peritoneal wash-out.
A total of 65 exchanges or reinsertions (via the mature tract) were required within our institution. An exchange was never performed via an immature tract. Some patients required repeat exchanges or reinsertions. The reasons for exchange/reinsertion are outlined in Table 3. Patients were not actively followed up but rather if an exchange or second primary RIG was performed, it was recorded.
In 312 (64.7%) patients a 14 Fr button tube was inserted, an 18 Fr button tube in 164 (34%) patients, a 16 button Fr tube in four patients and a 12 Fr button tube was inserted in two patients. Tube size was left to the discretion of the operators.

Discussion
Gastrostomy insertion is the long-term gold standard alternative for enteral feeding in patients who cannot feed orally. Both RIG and PEG tube insertions are safe procedures with acceptable mortality and morbidity rate given the long-term nutritional benefits. [1][2][3] A large meta-analysis performed by Wollman et al. 1 in 1995 with 5680 cases showed significantly fewer complications in the RIG cohort when compared to the PEG cohort and also a significantly higher success rate (99.2% vs. 95.7%), although an Australian paper by Cherian et al. 3 showed that RIG was more likely to dislodge compared to PEG. Another meta-analysis published in 2020 by Yuan et al. 4 examined the safety profile, success rate and 30-day mortality associated with RIG versus PEG in patients with MND. Seven studies with a total of 603 patients were examined. The meta-analysis demonstrated a 90.15% success rate in PEG and 96.76% success rate in RIG, which was found to be a statistically significant difference strongly favouring RIG (OR = 3.96, 95% CI (1.31-12.02); P = 0.02). 4 This meta-analysis confirms the results of previous studies, carried out in this institute and others, regarding the risk of aspiration for MND patients at endoscopy as well as difficulty tolerating the scope and therefore resulting in higher success rates for RIG over PEG. [5][6][7] RIG is often the preferred choice in patients with large head, neck or oesophageal cancers where it is not mechanically possible to fit an endoscope past the obstructing mass and therefore not possible for PEG insertion. 8 Unfortunately, these comparative studies did not indicate the proportion, or indeed if any of the RIG's  not report any mortality directly related to the RIG insertion but a 30-day all-cause mortality of 6.7% was recorded. This study was comparable to ours in the technique and equipment used to insert the mushroom button; however, it focused only on MND patients and had a relatively small cohort of patients (n = 104). Shaw et al. 9 also published a paper assessing the feasibility of the 'Entristar skin-level' gastrostomy tube for MND patients in 25 patients which had a 100% success rate and an 8% major complication rate. Our institute is the national neurosurgical centre and therefore a large volume of the RIG referrals comes from this department (48.9% of cases we performed were in patients with neurological disorders). Referral pathway in our institute is heterogenous with the majority of the medical teams requesting PEG and surgical teams and neurology teams requesting RIGs, we acknowledge that different practices occur in different institutions. It is also acknowledged that this underrepresents the actual number, as many of the RIG's were replaced in the community once a mature tract was present. As previously discussed, traditionally, radiologically inserted 'button' gastrostomies are inserted in a two-step procedure; initially, a long gastrostomy is inserted and then at a later date a 'button' gastrostomy is inserted once a mature tract has developed; however, de novo insertion of button RIG had previously been described. 2,7,10,11 In 2013, Power et al. 2 reported a success rate of 95.3% for the insertion of a primary 'button' RIG and a 30-day allcause mortality of 6.8%. This study had a large cohort of 260 patients in whom major complications were noted in 1.2% and minor complications in 12.8%. 2 Lyon et al. 10 published similarly high rates of success at 98% for primary 'button' RIG insertion in a smaller cohort of 53 patients.
In our institution, we changed our RIG practice from standard RIG tubes to button tubes for a few reasons. First, due to confusion/disorientation of many neurosurgical patients it was found a significant number of patients inadvertently removed the 'longer' gastrostomy tubes prior to the planned insertion of the shorter 'button' gastrostomy. The low-profile external component of the 'button' makes it less prone to 'pulling' motions, especially in this patient cohort. Secondly, many of the neurosurgical patients were repatriated to peripheral hospitals, once the acute neurosurgical issue had resolved, and therefore had to be expatriated to our institute for the 'second step' insertion of the 'button' gastrostomy. Thirdly, our dietetic department had a preference for button tubes and finally, 'button' gastrostomies are inherently shorter tubes than other gastrostomies and therefore are less likely to occlude from feeds. 2 There are also a variety of diameters available on the market up to 20 Fr; we usually insert either a size 14 Fr or 18 Fr.
However, button RIGs are not suitable for all patients. For example, patients with a very large BMI (tract length >5 cm), those with high-lying stomachs and patients with strong abdominal wall musculature (due to longer tract length) require a longer standard gastrostomy tube. As stated, we had a 97.1% success rate for primary button RIG insertion, in the 2.9% of patients who did not have a button placed, half of these patients had a standard gastrostomy tube inserted instead. These patients then returned 6 weeks later for button RIG insertion if tract length allowed.
A total of 65 exchanges or reinsertions were made over the 8-year period. We acknowledge that this underrepresents the actual number, as many of the RIG's were replaced in the community by Public Health Nurses. An exchange or reinsertion was never performed through an immature tract (we wait for 6 weeks for the tract to mature) due to the increased risk of malposition within the peritoneal cavity. Although this may seem like a large number of exchanges, the long follow-up period of 8 years over which this study is carried out means that most patients will require an exchange regardless of the type of gastrostomy tube inserted. It is also worth noting that a number of patients had a high number of exchanges carried out (one patient had five exchanges performed).
The 30-day all-cause mortality was 1% (n = 5), with 0.2% (n = 1) directly related to the procedure. One patient died as a consequence of malposition of the button within the peritoneal cavity with subsequent injection of feed which resulted in peritonitis and sepsis. Confirmation, with AP and lateral projections, of balloon position within the stomach as well as injection of contrast material to ensure no intraperitoneal spillage reduces the risk of malposition. Patients also commence on sterile water of 40 mL/h for 4 h prior to feed, the patient will generally report pain or demonstrate vital sign changes in this period if there is tube malposition. This complication is rare but if it occurs can have fatal consequences. Patient selection is also important in RIG tube insertion and as such we visit the patients prior to the procedure to ensure they are suitable for the procedure. assessed only MND patients while our patients are a more heterogeneous cohort. Wollman et al. 1 reported a 30-day procedure related mortality rate of 0.3%. In a large multicentre survey from the United Kingdom published in 2012 by Lowe et al. 12 they reported a 1% 30day mortality rate and 4% mortality rate in the follow-up period (1-4 months post RIG).
In conclusion, the primary insertion of a 'button' gastrostomy is a safe and effective method of RIG insertion and negates the requirement for revisiting the IR suite at a later date for 'button' insertion. Our success, complication and associated mortality rates are well within the limits of previous publications and this is the largest study from a single centre to date. It is recognized that there is a learning process with the adaption of this method of RIG insertion, but the benefits include single visit to IR suite, reduced risk of inadvertent removal and the low-profile external component means improved cosmesis.

Funding information
This paper was not supported by any funding.

Ethical approval
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. For this type of study, formal consent is not required. For this type of study, consent for publication is not required.

Data availability statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.