| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Postgraduate in Children and Adolescent Health, Federal University of
Pernambuco, Ave. Professor Moraes Rego, 1235 Cidade University, Recife,
Pernambuco 50670-901, Brazil.
2 Professor Fernando Figueira Mother and Child Institute, Pernambuco,
Brazil.
Received March 1, 2007;
accepted after revision May 27, 2007.
Address correspondence to E. J. da Costa e Silva
(eduardojust{at}ig.com.br).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. We retrospectively performed a case series study of all the abdominal CT scans on children and adolescents found in our archives. Two radiologists separately evaluated each CT scan twice. The radiologists were separately asked to formulate the most probable diagnosis and to decide whether tumor calcification was present. The first evaluation was performed without the unenhanced phase and the second was done with both the unenhanced and the contrast-enhanced scans. The agreement between the two methods, and that between each method and the histopathologic results, were measured using kappa statistics. The sensitivity and specificity of each method for diagnosing the more frequent neoplasms were also measured. The sensitivity and specificity of the contrast-enhanced CT scans were assessed for detecting calcification without reference to the unenhanced scan.
RESULTS. A total of 131 CT scans were evaluated. The agreement
between diagnoses from the two methods was almost perfect for both
radiologists (
= 0.97 and 0.99). No statistically significant
difference was seen between the two methods and the histopathologic results.
The sensitivity and specificity of the two methods for the most frequent
neoplasms were similar. The evaluations without the unenhanced phase showed
good sensitivity and specificity for tumor calcifications.
CONCLUSION. CT protocols without the unenhanced phase are a viable alternative for evaluating abdominal neoplasms in children and adolescents.
Keywords: abdominal imaging • gastrointestinal imaging • helical CT • pediatric imaging • radiation safety
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Currently, about 11% of imaging studies performed in the United States are CT [4]. There was a 92% increase in the number of pediatric abdominal and pelvic CT examinations in one American hospital between 1996 and 1999 [5]. This has given rise to considerable concern about the procedure, primarily about excessive radiation. Children are alleged to be 10 times more sensitive to radiation than adults [6]. The cancer-related mortality risk due to radiation exposure among children is estimated to be approximately one in 1,500 for a head CT examination and one in 550 for an abdominal CT examination [5].
Various efforts have been made to achieve better CT protocols for children: performing faster scanning, using less anesthetic, reducing motion artifacts and, especially, reducing the radiation exposure [4, 6–13]. Most studies have relied on reducing kilovolts and adjusting milliamperes by patient weight.
Each phase of the CT protocol contributes to the radiation dose. Kalra et al. [14] and Donnelly et al. [10] speculated on the possibility of omitting the unenhanced images without loss of information and reducing the radiation dose. However, this has not been recommended by all authors. Riccabona [15] emphasized the need to perform the unenhanced phase because of its capacity to detect calcification and hemorrhage.
Creation of protocols with a limited number of phases, in accordance with the specific clinical situation, would be one way to reduce exposure. The mainstream pediatric radiology practice in most centers today is to perform abdominal CT examinations without unenhanced scans. However, this practice has not been universally adopted. In clinical practice it is still common to perform pediatric examinations with the unenhanced scans. Careful studies about the need for these scans could be helpful in determining general adherence to these protocols.
The main objective of our study was to evaluate the use of abdominal CT without the unenhanced phase for diagnosing tumor types in children with a known abdominal neoplasm, using histopathologic results as the gold standard. The sensitivity and specificity of this protocol for detecting tumor calcifications were also evaluated, using unenhanced CT scans as the gold standard.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We performed a case series study, including all the abdominal contrast-enhanced CT scans found in the medical records of patients with confirmed abdominal neoplasms who were at our institution between January 1994 and September 2005. Only pretreatment studies were included. Histopathologic reports were used as the reference standards for tumor diagnoses. CT was performed in different radiology clinics in the region, and consequently no uniform protocol can be described. However, each examination underwent prior checking to ensure diagnostic quality.
The study was conducted between December 2005 and September 2006. Two CT radiologists took part in the study. Radiologist A has 30 years of experience in pediatric radiology and is the head of radiology in the diagnostic imaging department of a large pediatric hospital. Radiologist B has 8 years of experience in pediatric radiology, seven of them specifically in pediatric oncology.
The radiologists separately reviewed all CT images. Only hard copies were reviewed. Each CT scan was reviewed twice. On the first occasion, unenhanced images were not presented (protocol A). After 2 months, the same CT images were again presented with unenhanced images included (protocol B). On each occasion, the radiologists were asked to determine the most probable diagnosis. They were also asked to decide whether tumor calcifications were present. The only information provided was the patient's age and sex and that a neoplasm was the final diagnosis.
Intraobserver agreement of protocols A and B was measured with respect to
concordance between protocol A or protocol B and histopathologic reports. To
evaluate the agreement, kappa values were calculated. The strength of the
agreement was defined as suggested by Landis and Koch: < 0, poor;
0–0.20, slight; 0.21–0.4, fair; 0.41–0.6, moderate;
0.61–0.8, substantial; and 0.81–1.0, almost perfect
[16]. The interobserver
agreement for the two methods (protocol A and protocol B) was also evaluated,
comparing the results from each radiologist.
The sensitivity and specificity for diagnosing the four most frequent types of neoplasm were also calculated for protocols A and B, using histopathologic reports as the gold standard. The accuracy of contrast-enhanced scans (protocol A) for detecting tumor calcification was determined using the unenhanced scans (protocol B) as the gold standard.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
For all patients, each radiologist separately suggested a single diagnosis
based on the protocol A review. After 2 months, each radiologist again
suggested a single diagnosis based on protocol B. The agreement between the
results from protocol A and those from protocol B was almost perfect for both
radiologists (for radiologist A, = 0.978; for radiologist B, =
0.99).
For radiologist A, analysis of the results from protocol A versus
histopathology showed substantial agreement ( = 0.75 [95% CI =
0.666–0.846]). No statistically significant difference was seen
regarding the agreement between protocol B and histopathologic reports for the
same radiologist ( = 0.741 [0.651–0.831]).
Radiologist B had substantial agreement between protocol A and
histopathologic reports ( = 0.8051 [0.724–0.885]). Although the
agreement was not completely perfect for protocol B versus histopathology, the
difference was not statistically significant ( = 0.813
[0.7336–0.893]).
The interobserver agreement for protocol A was almost perfect ( =
0.81 [0.73–0.89]), whereas the interobserver agreement for protocol B
was substantial ( = 0.79 [0.70–0.84]). No statistically
significant difference was seen in the interobserver agreement for either
method.
The sensitivity and specificity of protocols A and B for the most frequent neoplasms were similar (Tables 2 and 3). The sensitivities of protocol A for tumor calcification were 74% (radiologist A) and 77% (radiologist B). The specificity was 100% for both radiologists.
|
|
For the patients in whom tumor calcifications were not detected on protocol A but were present on protocol B, no discordance was noted between the methods with regard to the most probable type of tumor.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our results showed substantial agreement between the protocol B and protocol A results for the two radiologists. This finding by itself indicates that unenhanced CT images for evaluating abdominal neoplasms in children give no additional information that is relevant to the diagnosis. Also, no statistically significant differences were seen regarding the agreements between protocol A and histopathology reports and between protocol B and histopathology reports for either radiologist.
Both methods showed good interobserver agreement, with no statistically significant difference, thus indicating no difference in reproducibility. Furthermore, the sensitivity and specificity of protocol A and those of protocol B for the most frequent types of neoplasms were similar.
One possible benefit of unenhanced images is to provide a measurement of tumor enhancement. This criterion is useful in adults. Enhancement is used as the most important criterion in differentiating surgical from non-surgical cystic renal masses according to the Bosniak classification [20]. However, simple renal cysts are not common in children, and pediatric renal neoplasms usually do not resemble simple cysts [1, 15].
Another possible advantage of the unenhanced images would be detection of calcification. The presence and characteristics of tumor calcifications used to be important criteria in evaluating conventional radiographs [17]. However, CT provides more detailed information, such as the originating organ, dissemination pattern, and presence of distant metastases [1]. It is unlikely that the presence or pattern of calcification would change the evaluation of an abdominal neoplasm in a child. Nonetheless, the results from our series indicate that protocol A has good sensitivity for detecting calcification. Moreover, when calcifications were not detected on protocol A but were present on protocol B, no discord was seen between the two methods with regard to tumor type, thus indicating that calcification did not affect the diagnostic decision.
The main steps in the interpretation of a CT scan obtained to evaluate an abdominal mass are to confirm its presence (mass vs no mass), locate it (peritoneal, retroperitoneal, renal, hepatic location), identify characteristics of the mass that might indicate certain types of tumors (calcification, fat) and, finally, check for local or distant dissemination. The most likely diagnosis is formulated by correlating certain clinical information, such as age, sex, presentation symptoms, and laboratory results. It is common to make lists of possible diagnoses. However, it is desirable to decide on the most likely diagnosis to determine the correct treatment.
Wilms' tumor is one example of this. Currently, there are two approaches for treatment of Wilms' tumor, both with good overall success rates [21, 22]. The North American approach (National Wilms' Tumor Study, NWTS) recommends initial nephrectomy followed by adjuvant therapy [21], whereas the European proposal (International Society of Pediatric Oncology, SIOP) is to make a presumptive diagnosis on the basis of imaging findings, followed by chemotherapy before surgery [22]. In both approaches, the initial treatment is based on the CT diagnosis of Wilms' tumor. Histopathologic confirmation before the initial treatment is required only in doubtful cases.
Thus, imaging studies should be optimized to provide the best accuracy. Consequently, any change in CT protocols must avoid reducing the capacity of the studies for making a correct most-likely diagnosis.
Development of protocols with reduced radiation exposure should be encouraged [6, 10]. Some researchers have shown the viability of shortened CT scans for evaluating specific conditions. Gnanasambandam and Olsen [23] retrospectively reviewed abdominal CT scans in children with neoplasms of the upper abdomen searching for pelvic abnormalities. They found pelvic abnormalities that would not have affected clinical management in six patients (2.6%). Four CT scans (1.7%) presented findings that could have affected management. However, only one of these abnormalities was not found in a previous imaging study performed within 1 week of the CT examination. Those authors concluded that the inclusion of the pelvis in routine abdominal CT for primary malignant tumors in the abdomen is not necessary [23].
These findings are similar to those reported by Kalra et al. [24]. In that study, a large number of "extra" images unrelated to the objective of the CT scans were seen in abdominal examinations. These extra images provided additional findings in 3% of the scans, but only in 1% would those additional findings have affected management [24].
Our study had some limitations. Both radiologists were aware that all the CT scans were obtained in children with abdominal neoplasms that were subsequently confirmed by histopathology reports. In clinical practice, some congenital or inflammatory lesions could simulate tumors in imaging examinations. In areas at high risk of tuberculosis, it might be difficult to differentiate tuberculosis from abdominal lymphoma. That both radiologists knew that a neoplasm was present contributed to the high specificity of CT in diagnosing lymphoma. The lack of clinical information should also be taken into account. It is unusual for radiologists to interpret CT scans without having minimum clinical information. In the case of abdominal neoplasms, CT is usually preceded by sonography or conventional radiography. The results from such studies, including the presence or absence of calcifications, were not available to our reviewers. Furthermore, they had access only to the physical films. Using a workstation can facilitate detection of calcifications on protocol A through manipulation of the images. It may also allow access to reformatted images in different planes, which enables better definition of the originating organ. Another limitation was the different protocols used for the CT scans because they were performed in different clinics; protocol A and protocol B were not independent studies. This lack of independence may contribute to overestimating our results because of incorporation bias.
We conclude that protocol B showed no advantage over protocol A for the initial evaluation of pediatric abdominal neoplasms. No significant difference was seen between the results from the two techniques with regard to the final diagnoses based on histopathology results. Protocol A also showed good sensitivity and specificity for detecting tumor calcifications. The presence of undetected calcifications in a few cases using protocol A did not affect the suggested diagnosis.
Further studies of using contrast-enhanced CT and eliminating the unenhanced phase in other applications should be performed.
Acknowledgments
We thank Silvio Cavalcanti de Albuquerque and João Vicente Ribeiro
Neto (Instituto Materno-Infantil Professor Fernando Figueira) for reviewing
all images of this study. We also thank Francisco Pedrosa and Arli Pedrosa
(Centro de Oncologia e Hematologia
Pediátrica–Recife/Pernambuco-Brazil).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. J. Goske, K. E. Applegate, J. Boylan, P. F. Butler, M. J. Callahan, B. D. Coley, S. Farley, D. P. Frush, M. Hernanz-Schulman, D. Jaramillo, et al. The Image Gently Campaign: Working Together to Change Practice Am. J. Roentgenol., February 1, 2008; 190(2): 273 - 274. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
