| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, German Cancer Research Center, Im Neuenheimer Feld
280, D-69120 Heidelberg, Germany.
3 Department of Diagnostic Radiology, University Hospital of Wales, Health Park,
Cardiff CF4 4XN, United Kingdom.
4 Biomedical SpA., Servizio di Radiologia, Via Prà 1/b, 16157 Genova,
Italy.
5 Department of Radiology, Leiden University Medical Center, Albinusdreef 2,
2333 AA Leiden, The Netherlands.
6 Istituto di Scienze Radiologiche e Formazione dell'Immagine, Ospedale SS.
Annunziata, Via P. Valignani 66100, Chieti, Italy.
7 Department of Radiology, University Hospital Nijmegen-St. Radboud, 6500 HB
Nijmegen, The Netherlands.
8 Eberhardt Karls-Universität, Radiologische Universitätsklinik,
Abteilung für Radiologische Diagnostik, Hoppe-Seyler-Str. 3, 72076
Tübingen, Germany.
9 MRI Department, Middlesex Hospital, Mortimer St., London W1 N8AA, United
Kingdom.
10 Medizinische Fakultät der Humboldt-Universität, Institut für
Röntgendiagnostik Charité, Schumannstr. 20/21, D-10098 Berlin,
Germany.
11 Radiologia e Diagnostica per Immagini, Ospedale Generale S. Giovanni Calibita,
Fatebene Fratelli, Isola Tiberina 39, 00186 Rome, Italy.
12 Instituto di Radiologia, Ente Ospedaliero di Pisa, Via Roma 67, 56125 Pisa,
Italy.
13 Institut für Radiologische Diagnostik,
Ludwig-Maximilians-Universität, Klinikum Grosshadern, 81377 Munich,
Germany.
14 Röntgendiagnostik I, Georg-August-Universität, Robert Koch Str. 40,
D-37075 Göttingen, Germany.
15 Department of Diagnostic Radiology, University Hospital, 66421 Homburg/Saar,
Germany.
16 State University Hospital, Hanzeplein 1, P. O. Box 30.001, 9700 RB Groningen,
The Netherlands.
17 Department of Radiology, Northwick Park Hospital, Watford Rd., Harrow HA1 3UG,
United Kingdom.
18 Institut für Radiologie Medizinischen, Universität Lübeck,
Ratzeburger Allee 160, D-23538 Lübeck, Germany.
19 Worldwide Medical Affairs, Bracco Imaging SpA., Via E. Folli 50, 20134 Milano,
Italy.
Received April 24, 2002;
accepted after revision December 31, 2002.
Address correspondence to M. V. Knopp.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Gadobenate dimeglumine at 0.05, 0.1, or 0.2 mmol/kg of body weight or gadopentetate dimeglumine at 0.1 mmol/kg of body weight was administered by IV bolus injection to 189 patients with known or suspected breast cancer. Coronal three-dimensional T1-weighted gradient-echo images were acquired before and at 0, 2, 4, 6, and 8 min after the administration of the dose. Images were evaluated for lesion presence, location, size, morphology, enhancement pattern, conspicuity, and type. Lesion signal intensity-time curves were acquired, and lesion matching with on-site final diagnosis was performed. A determination of global lesion detection from unenhanced to contrast-enhanced and combined images was performed, and evaluations were made of the diagnostic accuracy for lesion detection and characterization. A full safety evaluation was conducted.
RESULTS. Significant dose-related increases in global lesion detection were noted for patients who recieved gadobenate dimeglumine (p < 0.04, all evaluations). The sensitivity for detection was comparable for 0.1 and 0.2 mmol/kg of gadobenate dimeglumine, and specificity was highest with the 0.1 mmol/kg dose. Higher detection scores and higher sensitivity values for lesion characterization were found for 0.1 mmol/kg of gadobenate dimeglumine compared with 0.1 mmol/kg of gadopentetate dimeglumine, although more variable specificity values were obtained. No differences in safety were observed, and no serious adverse events were reported.
CONCLUSION. Gadobenate dimeglumine is a capable diagnostic agent for MRI of the breast. Although preliminary, our results suggest that 0.1 mmol/kg of gadobenate dimeglumine may offer advantages over doses of 0.05 and 0.2 mmol/kg of gadobenate dimeglumine and 0.1 mmol/kg of gadopentetate dimeglumine for breast lesion detection and characterization.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Although mammography is currently the most widely used technique for the detection and localization of breast abnormalities, the technique has variable specificity because of overlap of tissue densities and limited contrast between malignant and benign tissues. Furthermore, some 10% of palpable cancers may go undetected on mammography [2], and both mammography and sonography can be difficult to interpret in patients with breast implants [3], dense breast parenchyma [4], or surgical scars or deformity [5]. Review articles have indicated that MRI may have a key role to play in the detection and diagnosis of breast cancer [6-14], and its potential as a screening tool for women with a high risk for breast cancer has been reported [13, 15-17]. Advantageous attributes of MRI are its high soft-tissue contrast, multiplanar sectioning allowing visualization of tissue close to breast implants or the chest wall, and the absence of ionizing radiation. Gadolinium-enhanced MRI has been shown to be a sensitive technique for the detection of breast cancer, although variable specificities have been reported [6, 12, 18-24].
Gadobenate dimeglumine (MultiHance, Bracco Imaging, Milan, Italy) is a gadolinium-based contrast agent approved in Europe and elsewhere for MRI of the central nervous system and liver whose T1 relaxivity in vivo (9.7 mmol-1 x sec-1) is roughly twice that of gadopentetate dimeglumine (Magnevist, Schering, Berlin, Germany) and other approved gadolinium agents because of the capacity of the gadobenate molecule for weak and transient interaction with serum albumin [25-27]. This albumin-mediated relaxation enhancement may result in advantages for gadobenate dimeglumine over gadopentetate dimeglumine and other approved agents in tumors with high concentrations of albumin, possibly permitting similar relaxation enhancement to be achieved with lower overall doses or, in poorly vascularized tumors, greater relaxation enhancement to be achieved with an equivalent dose.
The aims of our phase II clinical trial were to evaluate doses of 0.05, 0.1, and 0.2 mmol/kg of gadobenate dimeglumine for contrast-enhanced MRI of the breast to establish the most appropriate dose for subsequent phase III evaluation, and to compare the results obtained after gadobenate dimeglumine administration with those obtained after administration of 0.1 mmol/kg of gadopentetate dimeglumine.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Patients
After the approval of the local ethics committee at each study center, 189
women (mean age, 54.5 years; range, 26-81 years) were enrolled and received
either gadobenate dimeglumine or the active comparator, gadopentetate
dimeglumine. For inclusion in the study, patients had to be at least 18 years
old and had to have had a mammographic examination within 30 days of contrast
agent administration that revealed an abnormality highly suspected of being
cancer so that the patient was highly likely to undergo either core or
surgical biopsy or surgery. Patients were excluded from the study if they were
pregnant or lactating; had received another investigational drug within 30
days before MRI; had received any other contrast agent during the 48 hr before
contrast agent administration; had received radiation therapy, chemotherapy,
or anticancer hormonal therapy before contrast agent administration; or had
any medical condition or other circumstances that would significantly decrease
the chances of obtaining reliable data or of achieving the study objectives.
Patients with a history of hypersensitivity to any gadolinium chelate or who
were otherwise contraindicated for MRI were also excluded from the study.
Patients who were eligible for inclusion were required to give written
informed consent before enrollment.
Enrolled patients were randomly assigned to one of four treatment groups to receive either gadobenate dimeglumine at a dose of 0.05, 0.1, or 0.2 mmol/kg of body weight, or gadopentetate dimeglumine at a dose of 0.1 mmol/kg of body weight. The four treatment groups were comparable with respect to demographic characteristics: 48 patients (mean age, 52.6 ± 11.3 years old) received 0.05 mmol/kg of gadobenate dimeglumine, 47 patients (53.2 ± 12.0 years old) received 0.1 mmol/kg of gadobenate dimeglumine, 47 patients (57.5 ± 12.1 years old) received 0.2 mmol/kg of gadobenate dimeglumine, and 47 patients (54.9 ± 12.6 years old) received 0.1 mmol/kg of gadopentetate dimeglumine.
Enrolled patients were permitted to undergo fine-needle aspiration not more than 7 days before or, preferably, within 1 month after contrast agent administration. Surgery and biopsy were permitted from 24 hr to 1 month after the administration of the study agent and the 24-hr safety follow-up period.
Contrast Material
Commercially available formulations of 0.5 mol of gadobenate dimeglumine
and 0.5 mol of gadopentetate dimeglumine were used. Doses of 0.1, 0.2, and 0.4
mL/kg of body weight (corresponding to final doses of 0.05, 0.1, and 0.2
mmol/kg, respectively) of gadobenate dimeglumine were administered, and a dose
of 0.2 mL/kg of body weight of gadopentetate dimeglumine was administered. All
doses were administered IV at a rate of 2 mL/sec (bolus) by means of a power
injector and were followed by a standardized saline flush of 10-20 mL.
Image Acquisition
MRI of the breast was performed with the patient in the prone position
using a dedicated double breast coil and an approved MR system. The 14 MR
scanners used were 10 1.5-T systems (Siemens, Erlangen, Germany: Magnetom
Vision, nine centers, 154 patients; Magnetom SP4000, one center, one patient),
one 1.0-T system (Siemens Magnetom Harmony, nine patients), two 1.0-T systems
(Gyroscan NT, Philips, Eindhoven, The Netherlands, 15 patients), and one 0.5-T
system (Signa Contour, General Electric Medical Systems, Milwaukee, WI, 10
patients). To achieve adequate spatial and temporal resolution, each imager
was required to have a gradient of at least 15 mT/m2.
Imaging at all centers was performed in the coronal plane, and parameters
were selected to cover the entire breast with thin slices (
3 mm) with no
interslice gap. Three-dimensional T1-weighted spoiled gradient-echo images
were acquired before contrast agent administration and at 0, 2, 4, 6, and 8
min after bolus injection of the contrast agent. Contrast-enhanced imaging
started after contrast agent injection, at the end of the saline flush. The
imaging parameters differed among centers because of the different MR
equipment available but were nevertheless kept constant throughout each study.
Imaging was performed with fat and water in-phase, a TR of 13 msec or less, 1
excitation, a rectangular field of view of 36 cm or less, a matrix of 128
x 256 or more, a flip angle of 10-35°, an in-plane resolution of 2
mm2 or less, and a total scanning time of 120 sec or less. No
fat-suppression sequences were used; to eliminate the signal of fat, we
performed image subtraction. The unenhanced images were subtracted from the
contrast-enhanced images by the investigator at each study center on a
pixel-by-pixel basis. Maximum-intensity-projection reconstruction of
subtracted images was performed for off-site image assessment. All images were
recorded on study-dedicated optical disks.
Image Assessment
All images were evaluated independently by two off-site observers who were
not affiliated with any of the centers in the study and who were unaware of
all patient data (identity, medical history, clinical profile, laboratory
results, results of comparative imaging procedures) and of the dose and
identity of the contrast agent administered. Image evaluation was conducted in
two separate sessions in a dedicated image analysis facility. The first
session was dedicated to the separate evaluation of unenhanced and
contrast-enhanced image sets, each in a fully random manner. The second
session was dedicated to the combined assessment of unenhanced,
contrast-enhanced, and subtracted image sets. During the combined assessment,
an additional maximum-intensity-projection reconstruction of the 2-min
contrast-enhanced image was displayed on a separate workstation to facilitate
improved lesion detection. During this assessment, each observer also had the
opportunity to place up to three regions of interest (ROIs) per lesion for the
purpose of generating signal intensity-time curves.
During both the separate and the combined interpretation sessions, the relative positions of detected lesions were indicated on breast maps for later use in lesion matching for the assessment of diagnostic accuracy. Lesion matching was performed by a third off-site observer who was uninvolved in the trial and who played no part in the initial off-site assessment. The purpose was to match any lesions detected by either of the two off-site observers during the separate and combined assessments with lesions recorded on final diagnosis maps by the on-site investigators on the basis of clinical history and the results of other procedures (i.e., mammography, sonography, fine-needle aspiration, biopsy, surgery, and pathology). Only lesion maps were viewed during this session; to avoid bias, we did not display MRIs at any time. The information provided for each lesion included only location (breast segment) and size. Lesion matching took place for each off-site observer individually; matching between observers was not performed.
Efficacy Determinations
Each image set (unenhanced, contrast-enhanced, and subtracted) was
evaluated first for technical quality. If an observer considered that an image
set was compromised (e.g., because of motion artifacts), no further efficacy
assessments for that image set were performed by that observer. Image sets
considered to be technically adequate were evaluated both qualitatively and
quantitatively.
Qualitative efficacy evaluation.—A preliminary evaluation was performed to rate the change from unenhanced images to contrast-enhanced images and combined (unenhanced, contrast-enhanced, and subtracted) images in the off-site assessment of global lesion detection for each dose of contrast agent administered. For this evaluation, each off-site observer was required to assign a lesion detection score of 0, 1, or 2 to each lesion detected, with 0 = uncertain, 1 = possibly or probably present, and 2 = definitely present. A score of -1 was assigned retrospectively to lesions that were seen on the final diagnosis but were not detected on MRI by the off-site observer (i.e., false-negatives). Lesions detected on MRI that were not present at the time of the final diagnosis (i.e., false-positives) were scored as missing lesions. The global lesion detection score was determined as the average of the lesion detection scores for all "final diagnosis" lesions of a patient.
Additional qualitative assessments of efficacy were based on determinations of lesion size, lesion margins, lesion enhancement, lesion conspicuity, the morphology of lesion enhancement, and lesion type (Table 1). These determinations were performed using predefined assessment scales for all lesions detected on each image set (both separate and combined) if 10 or fewer lesions were detected. If more than 10 lesions were detected on a given image set, assessment was performed for only 10 lesions, starting with lesions considered to be malignant.
|
Quantitative efficacy evaluation.—Quantitative
determinations of lesion enhancement were performed during the second off-site
interpretation session, which was dedicated to the evaluation of combined
images. For these evaluations, up to three ROIs per lesion were placed to
generate signal intensity-time curves. The most representative curve for each
lesion was then described according to the signal intensity enhancement rate
(a 3-point scale in which 1 = slow, 2 = intermediate, and 3 = fast) and the
time course for signal intensity enhancement (a 3-point scale in which 1 =
steady increase, 2 = plateau, and 3 = washout). Quantitative assessments of
lesion enhancement were calculated using the signal intensity from the
unenhanced image (SIpre) and the maximum signal intensity from the
contrast-enhanced dynamic signal intensity curve (SIpost) according
to the following equation:
 |
Diagnostic accuracy.—Diagnostic accuracy was analyzed per lesion and per breast for both lesion detection and lesion characterization. For the determination of the diagnostic accuracy of lesion detection, lesions indicated on the off-site assessment maps for the unenhanced, contrast-enhanced, and combined MRI sets were compared in terms of size and location with lesions indicated on the final diagnosis reference maps prepared by the on-site investigators. The final diagnosis reference maps were identical in format to those used for the off-site assessments and were prepared on the basis of information available from other diagnostic procedures (i.e., mammography [mandatory], sonography, fine-needle aspiration, biopsy, surgery, or pathology). The sensitivity for lesion detection on a per lesion basis was defined as the proportion of final diagnosis lesions (pooled over all patients) that were detected on each of the separate and combined MRI sets. Specificity on a per lesion basis was not defined because it was not possible to designate a lesion as a true-negative.
In terms of the per breast analysis, a true-positive breast was one in which at least one lesion was detected on both MRI and at the final diagnosis. Conversely, a true-negative breast was one in which no lesions were detected either on MRI or on final diagnosis. If at least one lesion was detected in a breast on the final diagnosis but no lesions were detected on MRI, the breast was considered false-negative. Similarly, if no lesions were detected in a breast at the final diagnosis but at least one lesion was detected on MRI, the breast was considered false-positive. The sensitivity for lesion detection per breast was defined as the number of true-positive breasts / (total number of true-positive + total number of false-negative breasts). The specificity for lesion detection per breast was defined as the number of true-negative breasts / (total number of true-negative + total number of false-positive breasts).
Determination of the diagnostic accuracy for lesion characterization was performed only in patients with confirmed cytology or histology results available. Lesions were characterized on-site as either malignant or benign. Malignant lesions included noninvasive carcinoma (i.e., ductal or lobular carcinoma in situ) or invasive carcinoma (i.e., carcinomas that were invasive ductal, invasive ductal with extensive intraductal component, invasive lobular, mucinous, medullary, papillary, tubular, adenoid cystic, secretory [juvenile], apocrine, cribriform, Paget's disease of the nipple with or without invasive carcinoma, metaplastic carcinoma, inflammatory, or other). Benign lesions included duct ectasia, papilloma, adenosis, hyperplasia, atypical hyperplasia, fibroadenoma, and other.
Diagnostic accuracy for lesion characterization on a per lesion and a per breast basis was determined in a similar way to that described for lesion detection. Sensitivity on a per lesion basis was defined as the proportion of histologically confirmed malignant lesions (pooled over all patients) that were correctly classified as malignant on MRI. Specificity per lesion was defined as the proportion of histologically confirmed nonmalignant lesions that were correctly classified as nonmalignant on MRI. On a per breast basis, sensitivity was defined as the proportion of breasts with at least one malignant lesion on histology that had at least one malignant lesion detected on MRI. Specificity per breast was defined as the proportion of breasts with no malignant lesions on histology that had no malignant lesions detected on MRI. The definitions for true- or false-positive breasts and true- or false-negative breasts were as described in the preceding text but with the presence or absence of malignancy by histology taken as the reference standard rather than the final diagnosis. In these evaluations, lesions that were considered indeterminate on MRI were pooled with those considered malignant (i.e., true-positive lesions) for the assessment of accuracy for characterization.
Statistical Analysis
The changes in the global lesion detection score facilitated by the three
doses of gadobenate dimeglumine and the one dose of gadopentetate dimeglumine
were compared using analysis of covariance with the unenhanced score as
covariate (F test). The level of significance used was 5%. Adjusted means
(including 95% confidence intervals) and p values were determined for
the overall treatment effect and, in the case of gadobenate dimeglumine only,
for the linearity of contrast enhancement. Post hoc chi-square tests were
performed to compare sensitivities in lesion detection on both a per lesion
and a per breast basis among each dose of gadobenate dimeglumine and
gadopentetate dimeglumine. As a protection against inflation of the overall
alpha level, comparisons were performed only if the global test, including all
doses of gadobenate dimeglumine and gadopentetate dimeglumine, was significant
at the 0.05 level.
For lesion conspicuity, qualitative lesion enhancement, and the confidence in lesion characterization, cross-tabulations were used to assess the change in the distribution of scores by study agent and dose from unenhanced images to contrast-enhanced images and combined (unenhanced, contrast-enhanced, and subtracted) images. Quantitative lesion enhancement (percentage of increase from unenhanced) was summarized for each study agent and each dose by lesion type (malignant or benign) on the basis of histology results.
Agreement between the two off-site observers was assessed by evaluating correlation coefficients with 95% confidence intervals for global lesion detection and by means of kappa statistics with 95% confidence intervals for determinations of diagnostic accuracy for lesion detection and characterization. Each analysis was performed separately for each image set by study agent and dose.
Safety Evaluations
Clinical monitoring (physical examination, vital sign measurements) was
performed before the patient entered the bore of the magnet (predose),
immediately after the patient left the magnet (postdose), and 24 hr after the
MRI procedure (follow-up). Additionally, blood samples were obtained predose
and at the 24-hr follow-up examination. All safety data were reviewed by a
data monitor expert.
The safety profiles of the two agents were evaluated in terms of the incidence of clinical adverse events and in terms of changes from predose in physical examination, vital signs, and laboratory variables (hematology and blood chemistry). Adverse events were classified as either serious (i.e., death, life-threatening, requiring or prolonging hospitalization, or resulting in persistent or signifi-cant disability or incapacity) or nonserious (rated as mild, moderate, or severe). The relationship of each adverse event to the study agent was classified as probable, possible, not related, or unknown.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Lesion Detection
Global lesion detection.—Increases in global lesion
detection score over unenhanced images alone were noted by observers 1 and 2
for both the contrast-enhanced images and the combined images
(Fig. 1). For the
contrast-enhanced images, a statistically signifi-cant (p < 0.05)
increase in the global lesion detection score from the scores for the
unenhanced images was noted by both observers for all dose groups except the
0.05 mmol/kg of gadobenate dimeglumine group in the case of observer 1. For
the combined images, a statistically significant (p < 0.05)
increase was noted by both observers for all dose groups without exception.
Overall treatment effect was shown to be significant for both observers for
the increase from unenhanced to contrast-enhanced images (observer 1:
p = 0.009; observer 2: p = 0.01) but by observer 1 only for
the increase from unenhanced images to the combined images (observer 1:
p = 0.027; observer 2: p = 0.404).
|
For patients who recieved gadobenate dimeglumine, statistically significant increasing trends in the global lesion detection score with increasing dose were noted by both observers for both the contrast-enhanced image set alone (observer 1: p < 0.001; observer 2: p = 0.003) and the combined image set (observer 1: p = 0.003; observer 2: p = 0.035) when compared with unenhanced images.
Baseline adjusted means suggested that the differences in global lesion detection scores between the 0.1 and 0.2 mmol/kg dose groups were generally smaller than those between the 0.05 and 0.1 mmol/kg groups. As regards patients administered 0.1 mmol/kg of gadopentetate dimeglumine, the changes noted for this dose group were generally greater than those seen for the group administered 0.05 mmol/kg of gadobenate dimeglumine but less than those seen for the group administered 0.1 mmol/kg of gadobenate dimeglumine. This pattern of results was similar for both observers.
Analysis of interobserver agreement for the global lesion detection score revealed no apparent dose-related trends or differences between the gadobenate dimeglumine and gadopentetate dimeglumine groups; correlation coefficients of 0.74, 0.57, and 0.88 were noted for the unenhanced, contrast-enhanced, and combined image sets for the gadopentetate dimeglumine group. For the gadobenate dimeglumine 0.05, 0.1, and 0.2 mmol/kg dose groups, the corresponding correlation coefficients showed ranges of 0.66-0.76, 0.5-0.85, and 0.61-0.71, respectively.
Diagnostic accuracy for lesion detection.—For all observers, image sets, and treatment groups, more patients had an increase than had a decrease in the number of lesions detected on contrast-enhanced images compared with unenhanced images. The largest percentages of patients with increases were seen in the 0.2 mmol/kg of gadobenate dimeglumine group; and for both observers, the increase in the number of lesions detected was greater for the combined image set than for the contrast-enhanced image set alone. No major differences were seen among treatment groups concerning the size of lesions detected on MRI. On combined image evaluation, most lesions for both observers were indicated to be between 1 and 5 cm in size.
Per lesion detection.—On the basis of the on-site final diagnosis, the total numbers of lesions present in the gadobenate dimeglumine 0.05, 0.1, and 0.2 mmol/kg dose groups and the gadopentetate dimeglumine 0.1 mmol/kg dose group were 85, 55, 79, and 82, respectively, for observer 1, and 84, 55, 79, and 82, respectively, for observer 2. The difference between observers in the number of lesions present in the gadobenate dimeglumine 0.05 mmol/kg dose group resulted from there being one fewer patient in this group in the case of observer 2. The sensitivity of MRI for the detection of lesions on a per lesion basis used the numbers of lesions at final diagnosis as reference standard (Fig. 2). On a per lesion basis, the greatest increases in sensitivity over unenhanced images alone were noted by both observers for the assessment of combined image sets in the 0.2 mmol/kg of gadobenate dimeglumine group (increases in sensitivity of an additional 40.5% and 39.2%, observers 1 and 2, respectively). This dose group, however, had the lowest sensitivity for lesion detection when unenhanced images alone were evaluated (24.1% [19/79 confirmed lesions] and 22.8% [18/79 confirmed lesions], observers 1 and 2, respectively). The highest overall sensitivities for lesion detection on a per lesion basis were noted for combined image sets in the 0.1 mmol/kg of gadobenate dimeglumine group (89.1% [49/55 confirmed lesions] and 74.5% [41/55 confirmed lesions], observers 1 and 2, respectively). The assessment of combined image sets of patients in the 0.1 mmol/kg of gadopentetate dimeglumine dose group revealed overall sensitivities of 62.2% (51/82 confirmed lesions) and 56.1% (46/82 confirmed lesions) for observers 1 and 2, respectively.
|
Statistical comparisons among dose groups were performed only when a global test of sensitivity involving all four treatment groups had initially shown significance. For the combined image sets, global testing revealed statistical significance for observer 1 (p = 0.003) but not for observer 2 (p = 0.176). Among-groups testing of combined images was therefore performed for observer 1 only. For this observer, the sensitivity observed on combined images for the gadopentetate dimeglumine 0.1 mmol/kg treatment group was significantly less than that observed for the gadobenate dimeglumine 0.1 mmol/kg group (p = 0.001). A similar evaluation was also performed for contrast-enhanced images only, in the absence of unenhanced and subtracted images. For these evaluations, global testing revealed statistical significance for both observers (observer 1: p = 0.008; observer 2: p = 0.034). Subsequent among-groups testing revealed significantly greater sensitivity for the gadobenate dimeglumine 0.1 mmol/kg treatment group than for the gadopentetate dimeglumine 0.1 mmol/kg treatment group for both observers (observer 1: p = 0.001; observer 2: p = 0.011).
An evaluation of the total number of false-positive interpretations (i.e., "lesions" detected by MRI that were not present at final diagnosis) revealed increased numbers on contrast-enhanced images compared with unenhanced images for all dose groups (Fig. 3). The most false-positive interpretations were noted by both observers during the assessment of combined image sets for the 0.2 mmol/kg of gadobenate dimeglumine group; however, little apparent agreement occurred between observers (79 false-positive interpretations for observer 1 compared with 37 for observer 2). The fewest false-positive interpretations were noted by both observers for the 0.1 mmol/kg of gadobenate dimeglumine group.
|
Per breast detection.—The numbers of breasts with lesions at final diagnosis for the gadobenate dimeglumine 0.05, 0.1, and 0.2 mmol/kg groups and the gadopentetate dimeglumine 0.1 mmol/kg group were 56, 47, 52, and 55, respectively, for observer 1, and 55, 47, 52, and 55, respectively, for observer 2. Again, the greatest increase in sensitivity over unenhanced imaging alone was observed by both observers for the combined assessment of images in the gadobenate dimeglumine 0.2 mmol/kg treatment group, for which fewer lesions were detected on unenhanced images (unenhanced sensitivities of 36.5% and 40.4% for observers 1 and 2, respectively, with increases in sensitivity of an additional 52.0% and 50.0%, respectively, for the assessment of combined images [Fig. 4]). The greatest overall sensitivities were noted for combined image assessment in the gadobenate dimeglumine 0.1 and 0.2 mmol/kg treatment groups (89.4% and 85.1% for the 0.1 mmol/kg group and 88.5% and 90.4% for the 0.2 mmol/kg group; observers 1 and 2, respectively). The per breast sensitivity values for the combined assessment of images in the gadopentetate dimeglumine 0.1 mmol/kg group were lower than those for the gadobenate dimeglumine 0.1 and 0.2 mmol/kg groups and similar to those for the gadobenate dimeglumine 0.05 mmol/kg group.
|
An evaluation of the specificity of lesion detection was possible on a per breast basis (Fig. 5). For this evaluation, specificity was defined as the percentage of breasts with no lesions at final diagnosis that had no lesions on MRI. Observer 1 recorded 40, 45, and 40 breasts in the gadobenate dimeglumine 0.05, 0.1, and 0.2 mmol/kg treatment groups, respectively, and 39 breasts in the gadopentetate dimeglumine 0.1 mmol/kg treatment group in which no lesions were detected at final diagnosis. The corresponding numbers for observer 2 were 39, 45, 40, and 39, respectively.
|
For both observers and all treatment groups, the specificity for lesion detection per breast on unenhanced images alone was high, between 90.0% and 95.6% for observer 1 and between 87.2% and 93.3% for observer 2. Although slightly reduced specificity compared with unenhanced imaging was noted on combined image assessment for the gadopentetate dimeglumine 0.1 mmol/kg group (observer 1: 67.5%; observer 2: 76.9%) and the gadobenate dimeglumine 0.2 mmol/kg group (observer 1: 67.5%; observer 2: 75.0%), less marked reductions in specificity were noted for the gadobenate dimeglumine 0.05 and 0.1 mmol/kg treatment groups by observer 1 (85.0% and 80.0%, respectively); in the case of observer 2, slightly improved specificity was noted (89.7% and 95.6%, respectively).
High levels of agreement for the detection or nondetection of lesions were
noted between observers for all treatment groups and image sets, particularly
for breasts with lesions detected at final diagnosis, for which the 95%
confidence intervals of the evaluated kappa values confirmed in all cases that
the agreement was unlikely to be caused by chance (
= 0.52, 0.73, 0.58,
and 0.34 for contrast-enhanced image sets alone; and = 0.64, 0.43,
0.90, and 0.75 for the combined image sets for the gadobenate dimeglumine
0.05, 0.1, and 0.2 mmol/kg treatment groups and the gadopentetate dimeglumine
0.1 mmol/kg treatment group, respectively). For breasts with no lesions
detected at final diagnosis, however, the 95% confidence intervals of the
evaluated kappa values confirmed that agreement was unlikely to be caused by
chance for the gadobenate dimeglumine 0.2 mmol/kg and gadopentetate
dimeglumine 0.1 mmol/kg treatment groups only ( = 0.21, -0.04, 0.29,
and 0.13 for contrast-enhanced image sets alone; and = 0.32, 0.12,
0.70, and 0.42 for combined image sets for the gadobenate dimeglumine 0.05,
0.1, and 0.2 mmol/kg treatment groups and the gadopentetate dimeglumine 0.1
mmol/kg treatment group, respectively).
Lesion conspicuity.—The mean lesion conspicuity scores per treatment group for unenhanced images alone ranged between 3.53 and 4.00 (intermediate to good) for observer 1 and between 2.65 and 2.96 (poor to intermediate) for observer 2. Despite the apparent differences in opinion, marked improvements in lesion conspicuity were noted after contrast administration by both observers for each treatment group and image set with the exception of the contrast-enhanced images alone for the 0.05 mmol/kg of gadobenate dimeglumine and 0.1 mmol/kg of gadopentetate dimeglumine groups of observer 1. A marked dose effect was noted for gadobenate dimeglumine for both contrast-enhanced images alone and combined image sets (conspicuity scores of 3.12, 4.07, and 4.23 and 4.05, 4.42, and 4.20, respectively, for observer 1, and conspicuity scores of 3.91, 4.13, and 4.30 and 4.51, 4.59, and 4.80, respectively, for observer 2 for the 0.05, 0.1, and 0.2 mmol/kg treatment groups, respectively). The mean conspicuity scores obtained after the administration of 0.1 mmol/kg of gadopentetate dimeglumine (2.85 and 3.62 for contrast-enhanced images alone and 4.31 and 4.38 for combined image sets, observers 1 and 2, respectively) were lower than those obtained with 0.1 mmol/kg of gadobenate dimeglumine.
During the assessment of combined images, observer 1 considered that conspicuity was improved for 44.7%, 60.6%, and 45.0% of lesions in the gadobenate dimeglumine 0.05, 0.1, and 0.2 mmol/kg treatment groups, respectively, and for 53.8% of lesions in the gadopentetate dimeglumine 0.1 mmol/kg treatment group. The corresponding numbers for observer 2 were higher because of the lower unenhanced scores (82.1%, 88.9%, and 95.0% for the gadobenate dimeglumine 0.05, 0.1, and 0.2 mmol/kg treatment groups, respectively, and 76.2% for the gadopentetate dimeglumine 0.1 mmol/kg treatment group).
Lesion Characterization
Confidence in lesion characterization.—The confidence of the
two observers in their lesion characterization was improved on
contrast-enhanced images compared with unenhanced images for all treatment
groups and image sets. The increases in confidence were generally similar
between observers, with no evidence of a dose-related trend among the
gadobenate dimeglumine groups. In most cases, the change in confidence was
from low to moderate confi-dence on unenhanced images to moderate to high
confidence on contrast-enhanced images. For observer 1, the mean confidence
score on unenhanced images was in the range of 0.8-1.03, which increased to
1.4-1.59 for contrast-enhanced images alone and to 1.45-1.89 for combined
images. The corresponding scores for observer 2 were 0.59-0.96 for unenhanced
images, 1.16-1.5 for contrast-enhanced images alone, and 1.64-1.76 for the
assessment of combined images. The greatest increase in mean confidence score
from unenhanced to contrast-enhanced images alone was recorded by observer 1
for the gadobenate dimeglumine 0.1 mmol/kg treatment group and by observer 2
for the gadobenate dimeglumine 0.2 mmol/kg treatment group (mean confidence
score increases of 0.69 and 0.65, respectively). For the assessment of
combined images, the greatest increases in mean confidence score were noted by
both observers for the gadobenate dimeglumine 0.05 mmol/kg treatment group
(mean confidence score increases of 0.89 and 1.05, respectively). No obvious
differences were apparent between gadobenate dimeglumine and gadopentetate
dimeglumine either in terms of the actual confidence scores reported or for
the mean increases in confidence from unenhanced images alone.
Diagnostic accuracy of lesion characterization.—Histology results were available for 174 (93.0%) of 187 patients: 41, 43, and 44 patients in the gadobenate dimeglumine 0.05, 0.1, and 0.2 mmol/kg treatment groups, respectively, and 46 patients in the gadopentetate dimeglumine 0.1 mmol/kg treatment group. Lesions were classified as malignant in 35, 29, 36, and 37 patients per treatment group, respectively. A total of 253 lesions were assessed histologically: 72 (28.5%), 53 (20.9%), and 67 (26.5%) in the gadobenate dimeglumine 0.05, 0.1, and 0.2 mmol/kg treatment groups, respectively, and 61 (24.1%) in the gadopentetate dimeglumine 0.1 mmol/kg treatment group. Of the 253 lesions assessed, 189 (74.7%) were classified as malignant and 64 (25.3%) as benign. The distribution of the more frequent lesion types among treatment groups is shown in Table 2.
|
Because of a higher proportion of patients with solitary lesions in the 0.1 mmol/kg of gadobenate dimeglumine group, and because of a larger number of multiple lesions in the other treatment groups, fewer lesions were characterized histologically for the 0.1 mmol/kg of gadobenate dimeglumine group compared with the other treatment groups. Furthermore, a comparatively high proportion of lesions were benign in this group (38% for the 0.1 mmol/kg of gadobenate dimeglumine group compared with 19% and 25% for the 0.05 and 0.2 mmol/kg of gadobenate dimeglumine groups, respectively, and 21% for the 0.1 mmol/kg of gadopentetate dimeglumine group). An unequal distribution of specific lesion types was noted in particular for benign fibroadenoma, which occurred eight times in the 0.1 mmol/kg of gadobenate dimeglumine group, three times in each of the 0.05 and 0.2 mmol/kg of gadobenate dimeglumine groups, but no times in the 0.1 mmol/kg of gadopentetate dimeglumine group.
The sensitivity, specificity, and overall accuracy for the characterization of lesions on the combined image sets are reported for observers 1 and 2 in Table 3. As regards the sensitivity for the characterization of malignant lesions, the highest value was noted for the 0.1 mmol/kg of gadobenate dimeglumine group (81.8% for both observers) compared with all other treatment groups. The greatest discrepancy between observers was noted in the specificity for lesion characterization in the 0.1 mmol/kg of gadobenate dimeglumine treatment group (observer 1 correctly classified 15 [75%] of 20 histologically proven lesions as not malignant compared with only 10 [50%] of 20 lesions correctly classified by observer 2). The 10 benign lesions incorrectly classified as malignant (false-positive interpretations) by observer 2 included six cases of fibroadenoma, four of which were detected in a single patient. On the basis of the sensitivity and specificity combined, the greatest overall accuracy among patients given gadobenate dimeglumine was reported for the 0.1 mmol/kg group (79.2% and 69.8% for observers 1 and 2, respectively). These values compared with values of 72.1% for both observers for the 0.1 mmol/kg of gadopentetate dimeglumine group, for which the proportion of malignant lesions was comparatively high.
|
Evaluation of contrast-enhanced images alone revealed similar trends to those observed during the combined image assessment, although lower overall sensitivities and comparatively high specificities for lesion characterization were seen (Table 4). Comparison among treatment groups in terms of the sensitivity for lesion characterization revealed markedly higher sensitivities for the 0.1 mmol/kg of gadobenate dimeglumine treatment group compared with the 0.1 mmol/kg of gadopentetate dimeglumine treatment group (observer 1: 66.7% vs 47.9%; observer 2: 81.8% vs 54.2%) and roughly similar sensitivities for the 0.05 mmol/kg of gadobenate dimeglumine treatment group compared with the 0.1 mmol/kg of gadopentetate dimeglumine treatment group (observer 1: 46.6% vs 47.9%; observer 2: 47.4% vs 54.2%).
|
Sensitivity for detection and accuracy of characterization by type of histologically proven lesion are shown for the most frequently occurring malignant lesion types in Table 5. The most frequently occurring type of malignant lesion was invasive ductal carcinoma (Figs. 6A, 6B, 6C, 6D, 7A, 7B, 7C, 7D, 8A, 8B, 8C, 8D, 8E, and 8F). Of the 88 lesions histologically diagnosed as invasive ductal carcinoma in patients who were given gadobenate dimeglumine (all dose groups combined), 71 (80.7%) were detected by observer 1 and 68 (77.3%) by observer 2. Of these lesions, observers 1 and 2 successfully classified 62 (87.3%) and 64 (94.1%), respectively, as malignant. For the 0.1 mmol/kg of gadobenate dimeglumine dose group alone, 19 and 18 lesions of 20 histologically proven invasive ductal carcinomas were detected by observers 1 and 2, respectively, of which 17 (89.5%) and 18 (100%), respectively, were accurately characterized as malignant. In comparison, of 23 lesions histologically diagnosed as invasive ductal carcinoma in patients given 0.1 mmol/kg of gadopentetate dimeglumine, observers 1 and 2 detected only 16 (69.6%) and 14 (60.9%) lesions, respectively, of which 15 (93.8%) of 16 and 14 (100%) of 14 (observers 1 and 2, respectively) were successfully classified as malignant.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
For patients given gadobenate dimeglumine, a greater proportion of histologically diagnosed invasive ductal carcinomas were detected and successfully classified as malignant in the 0.1 mmol/kg dose group than in the 0.05 and 0.2 mmol/kg dose groups. Similar findings in terms of the effect of the dose of gadobenate dimeglumine on the rate of lesion detection and characterization were obtained for lesions histologically diagnosed as invasive ductal carcinoma with an extensive intraductal component and invasive lobular carcinoma. However, it is difficult to draw any meaningful conclusions about these and the other malignant lesion types because of the comparatively low frequency with which they occurred in each dose group. Similarly, the numbers of lesions histologically diagnosed as benign were too small for us to attach meaningful conclusions to the data as a whole, although detection of benign lesions with contrast uptake was in all cases 100% by both observers.
Lesion Enhancement
Qualitative lesion enhancement was noted for most lesions in all treatment
groups after contrast agent administration. Evaluation of contrast-enhanced
images alone revealed lesion enhancement for at least 80% (observer 1) and 86%
(observer 2) of all lesions detected on unenhanced and contrast-enhanced
images in each treatment group. Similar results were obtained for the
evaluation of the combined image set.
As regards the enhancement patterns of the different lesion types, no obvious differences were apparent among the different treatment groups. In terms of lesion morphology, most malignant lesion types in each treatment group were categorized by observer 1 as having "rim enhancement in irregularly shaped lesion." Observer 2 likewise classified a high proportion of lesions in each treatment group in a similar way. In the case of lesion margins, observer 2 classified those of most lesions in each treatment group as being "spiculated." A greater variety of responses was obtained from observer 1 for the evaluation of lesion margins but because of the comparatively low numbers of lesions of different types in each group, no intergroup differences could be discerned and no meaningful conclusions drawn. However, for invasive ductal carcinoma, which was the most frequently occurring lesion type in each of the four treatment groups, observer 1 classified the margins as being either "lobulated" or "irregular or ill-defined" for most lesions in each group.
In terms of the rate of lesion enhancement, no differences were discernible among treatment groups or lesion types by either observer. Overall, the enhancement rate was considered "fast" for at least 75% of all malignant lesions in each treatment group by both observers. Similarly, the signal intensity-time curves for most malignant lesions in each treatment group were classified as "washout" by both observers. Unfortunately, too few benign lesions of different types were present in each treatment group to adequately describe an overall signal intensity-time curve pattern for these lesions. Nevertheless, for the gadobenate dimeglumine 0.1 mmol/kg group for which the greatest number of benign lesions were present, the overall tendency was for the signal intensity of benign lesions to either increase steadily or plateau with time.
An evaluation of the quantitative enhancement of malignant lesions in each treatment group was performed. For observer 1, a total of 33, 38, 46, and 51 ROIs were placed on malignant lesions in the gadobenate dimeglumine 0.05, 0.1, and 0.2 mmol/kg dose groups and the gadopentetate dimeglumine 0.1 mmol/kg dose group, respectively. For observer 2, the corresponding numbers were 31, 25, 26, and 33 ROIs. Substantial increases in lesion enhancement compared with unenhanced images were noted by both observers for all treatment groups, with a marked effect of dose evident for gadobenate dimeglumine (Fig. 9). The lesion enhancement obtained with 0.1 mmol/kg of gadopentetate dimeglumine was slightly less than that obtained with 0.05 mmol/kg of gadobenate dimeglumine. Although a similar trend was indicated for benign lesions, too few ROIs (between four and 13 for observer 1 and between two and six for observer 2) were placed to draw any definite conclusions.
|
Safety
The overall incidence of patients experiencing adverse events was 12.7% for
patients treated with gadobenate dimeglumine and 14.9% for patients treated
with gadopentetate dimeglumine. The incidence of patients experiencing adverse
events considered to be of probable, possible, or unknown relation to the
study agent was 11.3% for patients treated with gadobenate dimeglumine and
10.6% for patients treated with gadopentetate dimeglumine. No evidence of any
dose-related trends in the patients treated with gadobenate dimeglumine was
found. Headache was the most frequently reported event (reported by one
patient [2.1%] in the 0.05 mmol/kg of gadobenate dimeglumine group and by
three patients [6.4%] in both the 0.1 mmol/kg of gadobenate dimeglumine and
0.1 mmol/kg of gadopentetate dimeglumine groups). No other individual event
was reported by more than two patients in any treatment group. No serious
adverse events were reported during the study, and no patient withdrew from
the study because of an adverse event. All adverse events were classified by
the investigator as either mild or moderate in intensity. No clinically
significant trends in vital signs or laboratory parameters were observed.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The principal finding from this initial study is that a dose of 0.1 mmol/kg of gadobenate dimeglumine may offer advantages over doses of 0.05 and 0.2 mmol/kg of gadobenate dimeglumine and 0.1 mmol/kg of gadopentetate dimeglumine for MRI of patients with suspected breast cancer. Although efficacy was shown for all three gadobenate dimeglumine doses as indicated by the mean changes in the global lesion detection score from unenhanced to contrast-enhanced image sets, a marked increase in the global lesion detection score for the combined image set was noted by both observers only on passing from a dose of 0.05 mmol/kg of gadobenate dimeglumine to a dose of 0.1 mmol/kg of gadobenate dimeglumine. Only a negligible increase in the global lesion detection score was noted by observer 2 on passing from a dose of 0.1 mmol/kg of gadobenate dimeglumine to a dose of 0.2 mmol/kg of gadobenate dimeglumine.
Evaluation of the diagnostic accuracy of lesion detection also revealed a preference for a 0.1 mmol/kg dose of gadobenate dimeglumine; not only was the highest overall sensitivity for lesion detection recorded by both observers for both the contrast-enhanced only and the combined images with 0.1 mmol/kg of gadobenate dimeglumine, but this dose group was also associated with the fewest overall false-positive interpretations. In terms of the numbers of patients with additional lesions on contrast-enhanced images, most were noted in the 0.2 mmol/kg of gadobenate dimeglumine dose group. However, this dose group was also associated with the most false-positive interpretations. Furthermore, this group included a comparatively high number of patients with multifocal lesions in whom additional lesions are likely to be seen on contrast-enhanced images. In comparison, most patients in the 0.1 mmol/kg of gadobenate dimeglumine group had only a solitary lesion detected at final diagnosis. Not surprisingly, a consequence of the comparatively high numbers of false-positive findings reported for enhanced images compared with unenhanced images, particularly by observer 1 for the 0.2 mmol/kg of gadobenate dimeglumine dose group, was a reduced overall specificity on contrast-enhanced and combined images for lesion detection on a per breast basis.
A somewhat surprising finding in our study was the comparatively high numbers of lesions detected on unenhanced images alone. However, this finding can be attributed in large part to the inclusion criterion of the study that required the enrollment of patients with mammographic abnormalities suspected to be cancer who were highly likely to require histologic verification. Patients fulfilling this criterion probably had clearly identifiable lesions on unenhanced images alone and the blinded observers probably expected to find lesions. In this regard, the situation in a controlled clinical trial of this type is different from the situation in routine clinical practice, when not all patients undergoing MRI of the breast are subsequently found to have lesions. That the highest sensitivity for lesion detection and the lowest frequency of false-positive results on unenhanced images alone were noted for the 0.1 mmol/kg of gadobenate dimeglumine group can be attributed to a combination of the expectation of the blinded observers of finding lesions and the fact that most patients in this group had solitary rather than multifocal lesions as found in the other dose groups. Although the relatively high sensitivity for lesion detection on unenhanced images in the 0.1 mmol/kg
