Gastrostomy is a commonly performed procedure to provide enteral nutrition in malnourished patients or at risk thereof. Clinical practice and techniques currently vary widely worldwide.
The success of PEG, conventional PRG, or CT-guided gastrostomy is often dependent on the presence of upper digestive tract sufficient caliber to allow passage of the endoscope, a nasogastric tube, or at least a guide wire-catheter assembly to provide access for optimal gastric inflation.
In our study, a modified PRG technique was attempted in which the ultrasound probe was used to localize the collapsed stomach and to guide the gastric access without the use of nasogastric tube.
Pugash and coworkers  reported on the feasibility of using ultrasound as an alternative mean for gaining access to the stomach after using effervescent to identify gastric bubble. DeBaere et al.  described a similar technique using sodium bicarbonate powder as an effervescent. Subsequently published studies advocated US for rapid puncture after filling the stomach with water through an NG tube which has no technical advantage over an NG tube inserted for air inflation [13, 14]. Thus, the single use of ultrasound has not become the standard technique in patients with upper digestive tract obstruction.
Quadri et al.  succeeded to undertake PRG safely in 9 (100%) patients with complete obstruction of the upper digestive tract after failure of the conventional method. In 7 of the 9 patients (78%), initial gastric puncture was achieved under direct US visualization. In the remaining 2 patients, gastric puncture was achieved under fluoroscopy due to presence of locules of gas in the stomach.
Chan and coworkers  described also a modified percutaneous radiologic gastrostomy technique in 14 patients with malignant pharyngeal (13 patients) and esophageal (1 patient) complete obstruction. They were unable to obtain PEG or the nasogastric tube for the conventional PRG. They suspected the position of the collapsed stomach by the presence of any loculated air at the left subdiaphragmatic region. They punctured the presumed collapsed stomach using a 21G under fluoroscopic guidance. Then, the needle was withdrawn gradually with continuous water-soluble contrast medium injection under fluoroscopic control until the intragastic position of the needle tip was confirmed by delineation of the gastric rugae by the contrast. This technique had high risk of retroperitoneal structures injury and peritoneal contrast leakage. To avoid this blind gastric access, we adopted the technique reported by Quadri et al.  and we used the linear ultrasound probe to localize the collapsed stomach and to guide the gastric access. This allowed us to avoid posterior gastric wall transgression, retroperitoneal structures injury, and contrast peritoneal leakage.
Chan and co-workers  used ultrasound only to delineate the margin of the left hepatic lobe, which is marked on the skin to avoid its injury. We did not perform this step as we actually used the ultrasound as a main guidance for the gastric access. When we confirmed by ultrasound that the needle was within the stomach, we continued the procedure as described by Chan and co-workers using the fluoroscopy to guide the remaining steps of the technique.
Ultrasound-guided access to the stomach had been previously described. Lorentzen et al.  evaluate the effectiveness and safety of PRG under US and fluoroscopic guidance provided the stomach can be properly distended with fluid through a nasogastric tube. The fluid-filled stomach was punctured under US guidance. Schlottmann and colleagues  evaluate the use of transabdominal ultrasound for PEG placement in patients in whom transillumination failed. Gastrostomies were satisfactory in 14 of 15 cases (93.3%), and the complication rate was 28.5%.
Wu and coworkers  investigate the utility of bedside US during replacement of malfunctioning or dislodged G-tube through the previously fashioned tract. After insertion, color Doppler was applied over the catheter tip to enhance visualization during gentle tube oscillation
Heberlein and coworkers  reported successful placement of the gastrostomy tube in 82 (96.5%) out of the 85 patients without the use of nasogastric tube. Twenty-four patients had adequately distended stomach allowing fluoroscopic guided direct gastric puncture. The other 61 patients were given effervescent granules to distend the stomach before the puncture. The 3.5% failure rate was due to inadequate gastric distension by the effervescent granules.
Inaba et al.  2013 retrospectively investigated the medical records of 105 patients who underwent PRG. The technique used for PRG comprised insufflation to dilate the stomach via a nasogastric tube, followed by fluoroscopically guided puncture and gastrostomy tube placement. In patients for whom a nasogastric tube could not be inserted, the stomach was punctured with a fine needle under ultrasound-guidance and insufflated via this puncture needle to achieve dilation. The PRG procedure was successful in all cases including those who had undergone the modified technique.
Our technical success rate was 92% (23 out of 25 procedures) including 7 patients who had failed PEG and 5 patients who had failed PRG. This is similar to the success rates reported by Quadri et al.  and Chan et al. . Our success rate also falls within the results reported in previous studies using conventional percutaneous radiologic gastrostomy [12, 20, 21].
The 20% rate of minor complications in our study was also low approaching the minimum accepted rate per SIR and American Gastroenterology Association guidelines (13–43%) . Our rate of leakage (16%) was slightly higher than that with PEG but lower than that previously seen with RIG [23, 24]. No major complications were encountered. This is in contrast to the 6–7% rate of major complications reported by SIR .
Our study is limited by the small numbers included because of the low reference rate of patients with head and neck cancer with complete obstruction to our department during the study period. Furthermore, we did not compare the modified technique in our study to a group of patients to whom conventional technique was underwent. Further research in a randomized control fashion is warranted to optimize the procedure.