| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Department of Nuclear Medicine, Tuality Community Hospital, 335 SE 8th Avenue, Hillsboro, OR 97123. e-mail: gtkrishna@aol.com
Editor:
Since nine of 33 references cited by Dr Ziessman and colleagues (1) in their article in the November 2001 issue of Radiology were related to our work, we read their article with keen interest. We complement the authors for adding new information related to gallbladder function. In this communication, we would like to stress the importance of new information in their data, point out design flaws, and offer alternative explanations for some of their findings. When we introduced quantitative cholescintigraphy in 1981, we adopted 3-minute cholecystokinin (CCK)-8 infusion after finding out that 30- or 60-second bolus injection of 0.02 µg/kg, as recommended in the package insert for sincalide (Kinevac; Bracco Laboratories, Princeton, NJ), caused abnormal gallbladder emptying in many healthy subjects. We have used 3-minute infusion for all these years and found that the technique helps separate normal from abnormal gallbladder function reliably (2). On the basis of 3-minute infusion of CCK-8, in many series in the literature, investigators report excellent surgical outcome in well over 370 patients (3–5). In 1992, Dr Ziessman and colleagues (6) studied 23 healthy subjects with both 3- and 30-minute CCK-8 infusion and stated that 30-minute infusion was "advantageous." Data from the same 23 subjects were revisited in 1999 (7). In the current study, they conclude with the statement, "infusions of 30–60 minutes are required" (1). Neither now nor in 1992 or 1999 do they provide any patient data to show that a longer duration of CCK-8 infusion is better than 3-minute infusion in separating normal from abnormal gallbladder function. This claim of "advantage" is contrary to excellent surgical outcome results obtained by several authors (2–4) in patients with chronic acalculous cholecystitis on the basis of 3-minute infusion.
The purpose of their study was to compare 3-minute with 60-minute infusion of CCK-8 and establish normal values. It was necessary then to control variables, which they did not. Infusion of 0.01 µg/kg of CCK-8 over 3 minutes results in a dose rate of 3.3 ng/kg/min, and infusion of the same dose over 60 minutes produces a dose rate of 0.17 ng/kg/min. An appropriate and pharmacologically acceptable method to compare one duration with the other would be infusion of an identical dose rate on both occasions, for 3 minutes and for 60 minutes. Only then would the authors be able to claim the advantage of one duration over the other.
We have found that gallbladder emptying does not follow a Gaussian type of response. When we used 3.3 ng/kg/min for 3 minutes or 3.0 ng/kg/min for 10 minutes, frequency distribution data were non-Gaussian, both in 75 healthy subjects and in 65 patients with chronic acalculous cholecystitis (unpublished data, 2002). Therefore, the use of the mean ± 2 SDs would be inappropriate to set the normal range. Instead, we use a cutoff level of 35% as a lower normal limit for 3-minute infusion and 50% for 10-minute infusion. Review of the data of Dr Ziessman and colleagues (1) in table 1 clearly shows that 15-, 30-, 45-, and 60-minute infusions also produce a non-Gaussian distribution. Since they do not have data in patients, a reliable cutoff level cannot be determined at this time.
Unlike the left ventricle, which has a fixed ejection fraction at a given time, gallbladder ejection fraction can be controlled to any desired level simply by changing the duration of CCK-8 infusion. This is clearly evident from the data of Dr Zeissman and colleagues (1) in table 1 and figure 2. Longer duration of infusion, as expected, simply shifts the mean value to the right. Establishment of upper or lower limit cutoff values requires frequency and/or distribution data in both healthy subjects and patients. In the absence of a complete supporting data set, it appears premature to suggest superiority of one method over the other, and authors should be cautious in making such categorical claims.
REFERENCES
Division of Nuclear Medicine, Georgetown University Hospital, 3800 Reservoir Road NW, Washington, DC 20007. e-mail: ziessmah@gunet.georgetown.edu
We appreciate the interest of Drs Krishnamurthy and Krishnamurthy in our article (1). We disagree with their assertion that there was a design flaw in our investigation, however, and do not accept their "alternative explanation" for our findings.
Drs Krishnamurthy and Krishnamurthy state that the results of our study did not show that a longer infusion is better than a 3-minute infusion in separating healthy subjects from patients with abnormal findings. The clearly stated purpose of our study was to establish normal values for both the 3- and 60-minute infusions of sincalide at 0.01 µg/kg. Normal values must be defined before the clinical utility of normal or abnormal values can be assessed. Our investigation was not a clinical study designed to assess the value of sincalide cholescintigraphy to diagnose chronic cholecystitis, nor was it a pharmacologic study designed to determine the optimal dose rate for sincalide.
It is inappropriate that clinical investigations be undertaken without first establishing normal values for a diagnostic technique. In this context, validation of the technique is as important as determination of whether a certain cutoff, say 35%, distinguishes normality from abnormality. For example, it is true that several investigators (1–3) found the lower limits of normality for gallbladder ejection fraction values to be between 30% and 40% by using slower (30–60-minute) infusions of sincalide. The problem is that this normal range cannot be reproduced with the short infusion method (1,2). The supraphysiologic serum levels of sincalide produced by the short infusion method (4) result in inadequate gallbladder contraction in 30%–35% of healthy subjects (1,2).
Drs Krishnamurthy and Krishnamurthy claim that over the years, they have found that a 3-minute infusion of sincalide "separates normal from abnormal gallbladder function reliably." The citation is merely a reference to their book, which includes no new data, and they have never published data to substantiate this claim, to our knowledge. They also state that by using a 3-minute infusion of sincalide, many investigators report excellent outcomes in the literature. Indeed, by using a variety of infusion methods and normal values, investigators have reported that sincalide cholescintigraphy is useful for confirming the diagnosis of chronic acalculous cholecystitis. It is important to note, however, that there are contrary data in at least three published investigations (5–7). Although the sum of the evidence may support the notion that sincalide infusion can confirm this diagnosis (3,8–11), an evidence-based review of this literature has yet to be performed. Many of the published study findings had small sample sizes and, with few exceptions, were retrospective, with the potential biases of this design. One possible exception is the study of Yap et al (3), a prospective randomized investigation with convincing results. It is important to note that they used a 45-minute infusion and studied 40 healthy control subjects to establish normal values.
An important caveat is that although the data in both of our investigations of the 3-minute infusion method (1,2) show a 30%–35% false-positive rate (low gallbladder ejection fraction) in healthy subjects, the false-positive rate might be lower in symptomatic patients. This is because healthy subjects have a low pretest probability of having gallbladder disease, while patients clinically referred for sincalide cholescintigraphy have a high pretest likelihood of gallbladder disease. These patients have typically undergone extensive work-up prior to referral to exclude other diseases and have been followed up for many months to years, allowing time for other causes to become manifest.
The Bayes theorem indicates that in healthy subjects who have a low pretest probability of disease, a positive test result is likely to be false-positive—that is, the positive predictive value is low. In patients with high pretest probability of disease, a negative test result is likely to be false-negative—that is, the negative predictive value is low. It is in those patients with an intermediate pretest probability for disease in whom both negative and positive results are most likely to be correct. It may be that an increased incidence of false-positive results is not noted because of the small sample size in most published studies. What our published data prove, however, is that false-positive gallbladder ejection fractions are much more likely to occur with 3-minute infusion than with 30–60-minute infusion. We think that is hard to argue with the assertion that false-positive rates should be as low as possible.
Drs Krishnamurthy and Krishnamurthy state that in their unpublished data, they found a non-Gaussian distribution of gallbladder ejection fraction values in healthy subjects and thus, our method of establishing normal values was not sta-tistically correct. However, the text in the Materials and Methods section of our article clearly indicates that we were aware of the statistical issues, and to determine normal values, we used methods not dependent on the Gaussian assumption.
We strongly believe that the data in our article (1), together with data in our prior article (2), are convincing evidence that a short 3-minute infusion of sincalide should not be used. Normal values cannot be established for the 3-minute infusion method because of the variability and wide range of gallbladder ejection fraction response in healthy subjects and the high false-positive rate. However, 30–60-minute infusions of 0.01–0.02 µg/kg of sincalide give results that are more physiologic, and normal values have been established for this technique, which has a low false-positive rate.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  G. T. Krishnamurthy, S. Krishnamurthy, and P. H. Brown Constancy and Variability of Gallbladder Ejection Fraction: Impact on Diagnosis and Therapy J. Nucl. Med., November 1, 2004; 45(11): 1872 - 1877. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
