The diagnostic efficacy of γ diagnostic peritoneal lavage. A, Sensitivity and specificity. B, Positive and negative predictive values, and accuracy. Independent 100% specificity and positive predictive value of the test. Sensitivity reaches a peak of 85% at 90 minutes without further improvement at 2 hours.
Gulec SA, Weintraub S, Wang Y, Cundiff J, Albarado R, Moulder P, O'Leary JP, Hunt JP. γ-Guided Diagnostic Peritoneal Lavage for Detection of Bowel Perforation. Arch Surg. 2004;139(10):1075-1078. doi:10.1001/archsurg.139.10.1075
Bowel perforation can be diagnosed by detection of orally administered technetium Tc 99m sulfur colloid (99mTc SC) in diagnostic peritoneal lavage (DPL) fluid using a handheld γ-detection probe.
A canine intestinal-injury model was used to test the hypothesis. The 99mTc SC (55.5 MBq) was administered in 500 mL of saline via a nasogastric tube. A DPL with 500 mL of saline was performed at 60, 90, and 120 minutes after administration of 99mTc SC. The radioactivity in the DPL effluent was measured using a handheld γ probe. A DPL effluent count that was 3 SDs above the background count was considered a positive test result. Twenty animals with perforation and 5 without perforation (negative control) were studied.
There were no false-positive γ-DPL test results. Sensitivity improved by time up to 90 minutes. The lowest positive count in the DPL effluent measured by the γ probe corresponded to 0.2% of the administered activity. No radioactivity was detected in blood and urine samples or liver and spleen specimens. The sensitivity, specificity, accuracy, positive predictive values, and negative predictive values at 90 minutes were 95%, 100%, 96%, 100%, and 83%, respectively.
γ-Guided DPL is a highly sensitive and 100% specific test in the detection of small-bowel perforation. Clinical studies are warranted to determine the patient-specific factors affecting diagnostic accuracy.
The incidence of bowel injury in blunt abdominal trauma is generally considered less than 1% but has been reported as high as 6%, and is approximately 15% in penetrating abdominal trauma.1- 3 The clinical and radiological detection of bowel perforations in blunt abdominal trauma is difficult. This difficulty poses a unique challenge in the era of nonoperative management of patients with abdominal trauma. Due to the lack of early objective physical evidence of intestinal tract injury and less than optimal radiological assessment, significant delays may occur in the diagnosis of intestinal tract injury. Diagnostic delays beyond 12 to 24 hours result in significant increases in morbidity and mortality.4,5
Computed tomography (CT) has evolved as a valuable diagnostic tool in evaluating stable patients with blunt abdominal trauma. However, overall sensitivity of CT in detecting bowel perforation is less than optimal. Pneumoperitoneum and extravasated oral contrast, the most specific findings of bowel injury, are rarely seen. Low-attenuation fluid between bowel loops, the most commonly seen finding, is often subtle and may easily be overlooked. The presence of a solid organ injury further decreases the diagnostic significance of free fluid in the abdominal cavity.6,7 There is a need for a more sensitive and specific test for diagnosis of bowel injury.
Technetium Tc 99m sulfur colloid (99mTc SC) is a radiopharmaceutical that has been safely used for many clinical applications, including gastric emptying (oral administration) and ventriculoperitoneal shunt patency (intraperitoneal administration) studies. We hypothesized that a break in the integrity of the gastrointestinal tract can be assessed by detection of orally (or nasogastric tube) administered 99mTc SC in DPL fluid using a handheld γ-detection probe at the bedside.
Twenty-five mongrel dogs (21-27 kg) were used for the study with the prior approval of the institutional animal care committee (IACUC#2027) at the Louisiana State University Health Sciences Center (New Orleans). Preprocedure housing and periprocedure handling of the animals were done following the institutional guidelines. Animals were kept overnight and were given nothing by mouth. Anesthesia was obtained using 20 to 30 mg/kg pentobarbital intravenously. An intravenous access was placed in the foreleg for the maintenance of the anesthesia. An endotracheal tube was inserted and the animals were allowed to breathe spontaneously. Heart rate and oxygen saturation were continuously monitored using a pulse oximeter. A 14F nasogastric tube was placed in the stomach for administration of the radiopharmaceutical. A femoral-vein cut-down and a urinary bladder catheterization were performed in 10 animals on which biodistribution data were collected.
A minilaparatomy was performed in the midline. The stomach, jejunum, and anatomical landmarks were identified and the position of the nasogastric tube was confirmed. In 20 animals, a single perforation measuring 1 cm in diameter was created in the proximal jejunum on the antimesenteric aspect at 20 cm distal to the ligament of Treitz. A 7F Jackson-Pratt catheter was placed in a dependent position in the abdominal cavity, and the proximal end of the catheter was brought out through a separate incision. The abdominal incision was closed in 2 layers in a running manner to avoid any possible leak. Five negative control animals underwent similar laparatomy and peritoneal catheter placement procedures, except no small-bowel perforations were created. Euthanasia was accomplished by administration of high-dose pentobarbital at the completion of the data collection at 2 hours after administration of the radiopharmaceutical. Postmortem liver and spleen tissue sampling was obtained in 5 animals in the experimental group to determine any radiolabeled sulfur colloid activity.
Fifty-five and one half MBq of 99mTc SC dispensed in 8 mL volume was administered via the nasogastric tube. The syringe was flushed with 10 mL of normal saline once and the nasogastric tube was flushed with 500 mL normal saline.
Peritoneal lavage using 500 mL of normal saline was initiated at 30 minutes after administration of the 99mTc SC, and repeated at 60, 90, and 120 minutes. A different lavage was performed at each time point. Successful lavage was defined as an instilled return of more than 75% of the fluid. Blood and urine samples (10 mL aliquots) were also collected simultaneously.
All samples were counted using a C-Trak handheld γ probe (Care Wise Inc, Morgan Hills, Calif). The γ peak centered on 140 keV with a 20% window. Ten-second counts were recorded for each specimen. Background counts were obtained before and during the procedure. Mean background activity was 2 ± 2 counts over 10 seconds. Activity in the DPL fluid was measured by placing the probe over the collection bag. A positive DPL was defined as a count level that was 3 SDs above the highest background count, which was calculated to be 10 counts in 10 seconds. Retention of the radiopharmaceutical in the syringe and in the nasogastric tube and distribution in the blood, urine, liver, and spleen were also measured using a dose calibrator. Total radiation exposure of the surgeon was measured using film badges.
The efficacy of the technique was tested by determining the sensitivity, specificity, accuracy, and the positive and negative predictive values.
Retention of the radiopharmaceutical in the syringe and in the nasogastric tube ranged from 33% to 41% and 0.5% to 5% of the labeled activity in the syringe, respectively. Mean activity that could be delivered to the stomach was 61.2%, due to the losses from the retention in the syringe and tubing. This reflected a 33.3 MBq stomach dose from an administered activity of 55.5 MBq (Table 1).
All lavages were successfully completed. The lowest γ probe counts considered to be positive in the peritoneal lavage return bag were 10 counts per 10 seconds (3 SDs above the background activity). This corresponded to an activity of 0.033 MBq in 500 mL. The detection sensitivity of the γ probe, therefore, was calculated to be 0.1% leakage of the intragastric activity into the peritoneal cavity.
Blood and urine samples tested negative for radioactivity at all time points, indicating the absence of free 99mTc leak into the systemic circulation. Liver and spleen samples also tested negative, confirming the absence of intact 99mTc SC leak into the circulation. There was no radioactivity (above background level) noted at any time point in the peritioneal lavage fluid in the negative control group (animals with no jejunal perforation).
Total radiation exposures to skin, eye, and deep body of the surgeon measured by a film badge were 1 mrem, 0.87 mrem, and 0.87 mrem, per procedure.
There were no false-positive results. True positivity improved by time up to 90 minutes. There was no further increase in accuracy beyond this point. The sensitivity, specificity, accuracy, positive predictive values, and negative predictive values at 90 minutes were 95%, 100%, 96%, 100%, and 83%, respectively. Overall results are shown in Figure 1.
Diagnostic peritoneal lavage was introduced in 1965,8 and since then there have been many studies9- 11 that report the clinical role of DPL, with or without CT scan, in abdominal trauma patients, as well as its role as an evaluation tool. In the majority of studies, a positive DPL was judged on the basis of high red blood cell counts. This typically reflects solid organ injury, which can frequently be managed without operation. No single DPL finding has been identified to be specific for bowel perforation. Once a very commonly used test, DPL is now used in a small subset of blunt abdominal trauma patients, namely those that are hemodynamically unstable and require rapid determination as to the presence of blood within the peritoneal cavity. Even this indication is being challenged by focused abdominal ultrasonography. Owing to the availability of more relatively noninvasive tests, such as CT and ultrasonography, the use of DPL has declined. The specific diagnosis of bowel perforation is extremely difficult to detect with DPL. Particulate material or bacteria in the lavage solution are rarely seen.1 Isolated white blood cell count elevation is also rare, and amylase and alkaline phosphatase levels are nonspecific for bowel perforation.12
A γ DPL with the radio tracer in the return fluid offers the advantage of detection of only extravasated intestinal content. A "hot" DPL return fluid indicates a breakdown in the bowel integrity with a specificity of 100%. The biodistribution of orally administered 99mTc SC proved to be confined to the gastrointestinal tract. No systemic absorption of intact 99mTc SC or free 99mTc was seen, as evident from the absence of uptake in the liver and spleen, as well as negative test results in blood and urine counts. A positive test result is defined as having a count rate that is more than 3 SDs above the background activity. With narrow γ probe settings (140-keV photo-peak with 20% window) the mean background activity was 2 ± 2 counts per 10 seconds. This level of background activity defines a positive lavage count at 10 counts per 10 seconds. By definition, positivity is dependent on the background counts and the γ-probe sensitivity, and therefore requires determination prior to individual diagnostic set up. The sensitivity of the γ DPL in this experimental model was found to be 95%. The data demonstrate increased sensitivity of the test up to 90 minutes, with no advantage in DPL performed thereafter. This remains within a reasonable time frame described for the evaluation of hemodynamically stable blunt-trauma patients.
Count rates in the DPL effluent ranged from 26 per 10 seconds to 2610 per 10 seconds. This range of difference in count rates is most likely due to the variability in the amount of the radiocolloid extravasation from the perforation site and the efficacy of the DPL fluid in mixing with extravasated radiocolloid activity. The extravasated fraction is dependent on multiple factors. Although different kinds of catheter material might affect the results by producing a surface binding of the radiocolloid, we have demonstrated that approximately 40% of the intended activity is lost up front by retention in the syringe and nasogastric tubing. Irrigation of the syringe and/or tubing has only a limited benefit. Gastric emptying and intestinal motility are also important factors in the delivery of the radiocolloid to the perforation site. More distal bowel perforations could possibly decrease the detection sensitivity. In the presence of trauma and injury, the propulsion of gastric and intestinal contents might be significantly hampered by the setting of ileus. The dispersion of the extravasated radiocolloid in the peritoneal space and actual physical contact with the DPL fluid is necessary to obtain a positive test result. In a clinical setting, this second factor may present a unique challenge. The optimal placement of a DPL catheter is not an uncommon problem. Similarly, complete return of the lavage fluid may be hindered by omental interference. It is also possible that different kinds of catheter material might affect the results by producing a surface binding of the radiocolloid.
Although the effects of pharmacologic interventions on the sensitivity of γ DPL were not studied in our experimental model, it could be argued that the use of prokinetic agents might increase the extravasation of the radiocolloid from the perforation site. Metoclopramide is known to stimulate the gastric emptying. Meglumine diatrizoate causes fluid secretion into the bowel due to its high osmotic load (1500 mOsm/L) and promotes small-bowel peristalsis. It has been estimated that the normal transit time from the small bowel to the colon is 30 to 60 minutes. After abdominal surgery, or in the presence of ileus, the contrast will reach the colon in 3 hours.13 In addition to the expected prokinetic effect of meglumine diatrizoate, there may be additional, practical advantages of combined administration of meglumine diatrizoate with the radiocolloid since it is routinely given to obtain an abdominal CT scan in a trauma setting. We also have demonstrated, in a separate set of experiments not included in this article, that the precoating of a nasogastric tube combined with post–radiocolloid-administration washing decreased the retention of the radiocolloid in the nasogastric tubing 10-fold. In a clinical setting the time required to complete a CT study would allow adequate lapse for the optimal timing for γ DPL assessment.
The intra-abdominal organ radiation dose estimates from oral and intraperitoneal administration of 37 MBq 99mTc SC have been well documented, as well as the exposure of medical personnel handling the patients. Direct measurements obtained during the γ DPL procedure demonstrated that the surgeon's exposure, despite being involved in every stage of the procedure (administration of the 99mTc SC, surgery, DPL effluent handling and disposal), was 1 mrem per procedure. This level of exposure is significantly less than the acceptable occupational exposure limits recommended by the Nuclear Regulatory Commission. However, from a clinical perspective, a positive γ DPL test result would require an operation in a contaminated field. We recognize this as a potential concern in clinical application. This concern is becoming much less prohibitive with the expanding applications of γ-probe–guided surgical techniques.14 Education of operating room and surgical personnel would be necessary.
γ-Guided DPL has been found to be a highly sensitive and 100% specific test in detection of small-bowel perforation in this experimental model. The potential problem with detecting more distal small-bowel lesions, particularly with delayed transit due to ileus and/or shock, was not addressed in this study. The potential for the test to decrease in sensitivity for more minimal injuries is also a possibility. Another drawback to our study design was the use of sequential lavages, which leaves fluid in the abdomen. This has the potential to alter the radiocolloid concentrations and thus effluent counts of the following lavage. The technique might have a role in the management of patients with both blunt and penetrating abdominal trauma. More basic and clinical studies are warranted to determine the patient-specific factors affecting diagnostic accuracy, as well as to determine a subset of abdominal trauma patients, possibly those with certain abdominal CT findings or clinical situations that could potentially benefit from this procedure.
Correspondence: John P. Hunt, MD, MPH, Department of Surgery, Louisiana State University Health Science Center, 1542 Tulane Ave, New Orleans, LA 70112 (firstname.lastname@example.org).
Accepted for publication February 18, 2004.