| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, University of North Carolina at Chapel Hill, Manning
Dr., Chapel Hill, NC 27599-7510.
2 Center for Excellence in Surgical Outcomes, Duke University Medical Center,
Durham, NC.
3 Department of Radiology, West Virginia University, Morgantown, WV.
4 Present address: Department of Radiology, Emory University, 1365 Clifton Rd.
NE, Bldg. A, Ste. 622, Atlanta, GA 30322.
5 Department of Radiology, University of Sao Paulo, Sao Paulo, Brazil.
6 Department of Surgery, Divisions of General Surgery and Surgical Research,
University Hospital Basel, Basel, Switzerland.
Received July 21, 2003;
accepted after revision October 30, 2003.
Address correspondence to R. C. Semelka
(richsem{at}med.unc.edu).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Sixteen breast cancer patients with liver metastases who underwent MRI before and after systemic therapy were retrospectively identified. On the basis of comparison of each MRI examination with the previous examination, disease status of the patients was classified as complete response, partial response, stable disease, or progressive disease. Liver metastases were characterized as hyper- or hypovascular on the basis of the degree of enhancement in the arterial, portal, and interstitial phases of imaging after administration of a contrast agent. Fisher's exact test and ordinal logistic regression models, including the type of systemic therapy, presence of multiple metastases, and hormone receptor status, were used to estimate the unadjusted and risk-adjusted association between the presence of hypervascular liver metastases and disease progression.
RESULTS. All patients in our sample (n = 16) were women and most (12/16, 75%) were white. Their median age was 51.5 years. In unadjusted analyses, the association between the presence of hypervascular liver metastases and disease progression was statistically significant (p < 0.0001). In multiple logistic regression analyses, hypervascular liver metastases were found to be an independent predictor of disease progression. Patients with hypervascular liver lesions were 20.5 times more likely to experience disease progression than patients without hypervascular metastases (odds ratio, 20.5; 95% confidence interval, 5.1–83.5; p < 0.0001).
CONCLUSION. Our analysis provides suggestive evidence that disease progression can be predicted through MRI assessment of the vascularity of liver metastases in patients with breast cancer.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MRI is an important diagnostic tool in the baseline and follow-up evaluation of primary and secondary lesions of the liver. Previous studies have shown that MRI has superior sensitivity and specificity to other imaging methods for the detection and characterization of metastatic liver lesions [9, 10].
The use of MRI to characterize the effect of systemic therapies on the vascularity of tumors has been previously described [11–13]. One report correlated a decrease in the size of a liver lesion after systemic therapy with decreasing tumor vascularity [11]. Another study reported a decreased enhancement of liver metastases in response to systemic chemotherapy in patients with metastatic colon cancer [12].
The correlation between an increase in the enhancement of liver parenchyma in the arterial dominant phase of dynamic CT and the subsequent development of metastases in cancer patients has been previously reported [14, 15]. However, to our knowledge, no study has shown an association between hypervascularity of liver metastases detected on MRI and disease progression. Although response status is an important predictor of patients' outcome [8], the identification of progression is particularly critical because that might result in a modification of the chemotherapy regimen [6]. The objective of our study was to evaluate the association of hypervascularity of liver metastases, as shown on MRI, and disease progression in breast cancer patients.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
One hundred three breast cancer patients were initially identified. Of these, 24 had no liver lesions, 34 had benign liver lesions, five had liver lesions that could not be characterized as benign or malignant on MRI, and 40 had liver metastases. Of the 40 patients with liver metastases, 16 met the inclusion criteria.
Patients were mostly white (12/16, 75%). All patients were women, and all had histologically proven invasive ductal carcinoma of the breast. Eleven patients (11/16, 68.8%) were lymph node–positive at the time of the breast cancer diagnosis. Estrogen and progesterone receptors were positive in nine (56.3%) of 16 patients and negative in five (31.3%). The tumor of one patient displayed positive estrogen and negative progesterone receptors, and in one patient the estrogen and progesterone status was unknown. Eleven patients (11/16, 68.8%) underwent chemotherapy, one underwent hormonal therapy (1/16, 6.3%), and four underwent combined chemo- and hormonal therapy (4/16, 25%). The median age of the 16 patients was 51.5 years (range, 36–78 years). Primary breast cancer was histologically confirmed in all patients. Liver metastases were detected after a median of 2 years (range, 0–13 years) after the diagnosis of breast cancer and were confirmed at liver biopsy in seven (43.8%) of 16 patients. In the remaining nine (56.3%) of 16 patients, a presumptive diagnosis of liver metastases was made on the basis of appearance on serial MRI examinations, clinical symptoms, and laboratory values.
All patients underwent systemic therapy based on different protocols. We thus divided the systemic treatment into three categories: chemotherapy, hormonal therapy, and a combination of both. All patients underwent MRI examination before and after treatment. The pretreatment MRI was performed within 1 month (median, 10 days; range, 1–30 days) of initiating systemic therapy and was considered the baseline examination. Ninety-nine MRI examinations were analyzed. Of these, six studies did not show perfect enhancement timing (see definition in section on Image Analysis) and were thus excluded from our analysis. Of the remaining 93 MRI examinations, 16 (baseline MRI studies) were performed before and 77 (follow-up MRI) were performed after the initiation of systemic therapy. The median number of MRI examinations performed per patient was 3.5 (range, 2–22).
The objective of this study was to assess whether the vascularity of liver metastases, as characterized on MRI, can be used to predict the disease status (complete response, partial response, stable disease, disease progression) as shown on follow-up MRI. Thus, the first MRI examination for each patient (the baseline MRI study, n = 16) provided only information regarding tumor vascularity but not regarding disease progression. Similarly, the last MRI examination of each patient was evaluated for disease progression only but not for tumor vascularity. Hence, although 93 MRI examinations were evaluated, a correlation between metastatic lesion vascularity and tumor response was determined in 77 examinations (93 minus 16).
All MRI studies were performed at the discretion of the attending oncologist and were adapted to the type and schedule of systemic treatment. Thus, time intervals between follow-up MRI after the initiation of systemic therapy were not standardized. The median interval between two MRI examinations was 70 days (range, 30–720 days).
MRI Technique
MRI of the upper abdomen was performed in 99 examinations using a 1.5-T MR
imager (Vision or Sonata, Siemens Medical Systems) with a phased array
multicoil for the body. MRI sequences included T2-weighted HASTE sequences
(TR/TE range, infinite/90–150) and T1-weighted breath-hold spoiled
gradient-echo sequences (TR range/TE, 130–150/4; flip angle, 80°;
slice thickness, 8–10 mm; 14–22 sections acquired in a 20-sec
breath-hold) before and after the administration of contrast material. In
these sequences, the section thickness was 7–10 mm with a 20%
intersectional gap and a matrix of 128–192 x 256 (phase frequency
encoding). A gadolinium chelate (Magnevist [gadopentetate dimeglumine],
Schering or Omniscan [gadodiamide], Amersham) was injected in a dose of 0.1
mmol/kg of body weight as a bolus injection at 2 mL/sec using a power injector
(Medrad). Spoiled gradient-echo sequences were repeated 18 sec (hepatic
arterial dominant phase) and 45 sec (portal phase) after contrast
administration. T1-weighted fat-suppressed breath-hold spoiled gradient-echo
images were acquired at 90 sec (interstitial phase) after contrast
administration.
Image Analysis
All MRI examinations were independently evaluated from hard copies by two
board-certified radiologists experienced in the interpretation of MRI.
Agreement was reached by consensus. To minimize expectation bias, the
investigators were unaware of the clinical course and laboratory parameters of
the patients. Serial MRI studies of individual patients were reviewed side by
side in chronologic order, starting with the images obtained before systemic
therapy. Each MRI study was compared with the previous study to define the
response status. Determination of lesion vascularity was made independently of
the interval change in lesion size.
According to World Health Organization criteria [16, 17], complete response was defined as the disappearance of all detectable liver lesions, and partial response was defined as a decrease of 50% or more in the size of a single liver lesion or as a decrease of 50% or more in the sum of the products of the perpendicular diameters of multiple lesions. The definition of stable disease was a decrease of less than 50% or an increase of less than 25% in total tumor size of one or multiple lesions. Finally, disease progression was defined as an increase of 25% or more in the size of one or multiple liver lesions or as the appearance of new liver lesions.
The size of metastases with oval or rounded morphology was calculated (using unenhanced T1-weighted images) by multiplying the longest diameter by the greatest perpendicular diameter in the transverse plane [16, 17].
Both lesional and perilesional areas were examined [12]. The degree of enhancement after administration of contrast material was evaluated on arterial, portal, and interstitial phases. To achieve a standardized analysis, we considered that perfect enhancement timing in the arterial dominant phase occurred when contrast material was detected in the aorta and the principal branches of the portal vein, but not in the hepatic veins. According to the degree of enhancement, metastases were considered to be hypervascular—exhibiting intense enhancement similar to that of the pancreas and the renal cortex in the arterial dominant phase and becoming less conspicuous in portal and interstitial phases—or hypovascular—exhibiting no enhancement or enhancement less than that of the pancreas or the renal cortex on the arterial dominant phase and becoming more conspicuous in the portal and interstitial phases.
If both hyper- and hypovascular metastases were found on the same MRI study, the patient was classified as having hypervascular liver lesions if most lesions were hypervascular. Similarly, the patient was classified as having hypovascular liver lesions if most lesions were hypovascular. Vascularity of liver metastases was reevaluated for each subsequent MRI examination.
Statistical Analysis
Descriptive analyses were performed using means and standard deviations for
continuous variables and frequencies and percentages for categoric variables.
Unadjusted comparisons of patients' baseline characteristics between the
subsets with (n = 10) and without (n = 6) hypervascular
liver lesions were performed using Mann-Whitney and Fisher's exact tests. The
unadjusted association between the presence of hypervascularity of liver
metastases and disease progression was assessed using Fisher's exact test. The
risk-adjusted association between degree of vascularization of the liver
metastases measured on MRI and disease progression status was evaluated using
ordinal logistic regression models
[18] and generalized
estimating equations with observations on metastatic lesions clustered in
patients [19]. Covariates in
the multivariable models included the presence of multiple liver metastases,
the hormone receptor status, and the type of systemic treatment received
(chemotherapy, hormonal therapy, or combination of both). Each MRI study was
considered to be a baseline for the next scan in terms of classifying the
disease status. Odds ratios and 95% confidence intervals (CIs) were computed
using both unadjusted and risk-adjusted generalized estimating equation
models. In addition to our primary end point (disease progression), we also
assessed the association between hypervascularity of liver metastases and a
composite end point (disease progression and stable disease). In other words,
all patients having either progressive disease or stable disease were coded as
having the event of interest.
Furthermore, because the number of MRI examinations after systemic treatment was not standardized in our investigation and varied considerably, we evaluated the potential impact of clustering of MRI examinations in patients in sensitivity analyses: patients with the most MRI examinations were removed from our statistical models one at a time. The risk-adjusted models were then rerun to assess whether patients with many MRI examinations distorted our findings.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
According to the findings of the first and the last MRI examinations, 12.5% (2/16) of patients were considered to have complete response, 12.5% (2/16) to have partial response, 31.3% (5/16) to have stable disease, and 43.8% (7/16) to have progressive disease.
The subset of patients who had disease progression (n = 7) was similar to patients who had stable or regressing disease (n = 9) with respect to age (p = 0.8) and race (p = 0.7). In unadjusted analyses, the association between the presence of hypervascularity of liver metastases and disease progression was statistically significant (p < 0.0001).
In multiple logistic regression analyses after adjusting for the type of systemic therapy, presence of multiple liver metastases, and hormone receptor status, hypervascularity was found to be an independent predictor of disease progression. On the basis of adjusted generalized estimating equations models predicting tumor progression, patients with hypervascular liver lesions were 20.5 times more likely to experience disease progression than patients without predominantly hypervascular lesions (odds ratio: 20.5; 95% CI, 5.1–83.5; p < 0.0001). Similarly, patients with hypervascular liver lesions were 17.4 times more likely to experience disease progression or stable disease than patients without predominantly hypervascular lesions (odds ratio, 17.4; 95% CI, 3.7–82.0; p < 0.0001) (Table 2). To assess the impact of potential clustering of MRI examinations in a patient, we performed sensitivity analyses in which patients with the most MRI studies were systematically removed (one at a time). Hypervascularity remained a significant predictor for disease progression even after removing the seven patients with the most MRI examinations, which indicates that clustering of MRI examinations in a patient did not threaten the validity of our findings. On the basis of adjusted ordinal logistic regression models, disease in patients with hypervascular lesions will progress in 62% of cases, will remain stable in 27% of cases, and will regress in 11% of cases.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Previous studies have shown an association of rapid increase of cancer vascularization, increase in tumor size, and the appearance of metastases [20–22]. Biologically, this phenomenon is related to angiogenesis [20–22]. Hypervascularity has been reported to reflect increased vascular density, increased mean vessel diameter, and greater vascular wall permeability [23–25]. Hypervascularity of liver lesions can also result from desmoplastic reactions, inflammatory cell infiltrates, and increased release of the vascular endothelial growth factor [12, 26, 27]. The major feature of hypervascular liver metastases on MR images acquired during the arterial dominant phase is an intense enhancement exceeding that of the liver parenchyma and similar to that of the pancreas and the renal cortex. During the portal and the interstitial phases, liver lesions become less conspicuous or obscured [28]. The pattern of lesional enhancement of hypervascular liver metastases can be shown as an intense ring enhancement, homogeneous enhancement, or heterogeneous enhancement.
In our patient sample, liver metastases from breast cancer were hypervascular in 10 patients (62.5%) and hypovascular in six (37.5%) before systemic therapy was begun. The predominance of hypervascular metastases from breast cancer before treatment is similar to that reported in previous studies [28, 29].
On evaluation after treatment, 35 (45.5%) of 77 MRI examinations showed hypervascular metastases, 40 (51.9%) of 77 showed hypovascular metastases, and two (2.6%) of 77 showed complete tumor regression. Systemic therapy has great importance in the treatment of liver metastases [7, 30]. Although the main underlying mechanism of action of these treatments is the inhibition of tumor cell proliferation, recent studies have shown that some systemic therapy agents also affect tumor angiogenesis. The inhibition of angiogenesis results in a decrease in vascularity, impaired tumor perfusion, and regression of tumor mass [20–22, 25, 31]. On MRI, the size and number of tumor-feeding blood vessels and the size of the tumor interstitium are important factors affecting the degree of lesion enhancement on gadolinium-enhanced images [32–34].
Hypovascular lesions in the arterial dominant phase of MRI show less intense enhancement than the pancreas and the renal cortex. Hypovascular lesions may have no enhancement, or equal or slightly greater enhancement than the liver parenchyma. Lesional enhancement commonly observed in the arterial phase in hypovascular tumors can be negligible or present as homogeneously isointense to hepatic parenchyma or as a faint ring enhancement. Independent of the pattern of enhancement, hypovascular metastases will generally be more conspicuous in the portal and the interstitial phases than in the arterial phase [28]. Perilesional enhancement usually has a wedge-shaped or circumferential morphology [12] and is more common in hypovascular metastases [12, 34].
In our patient sample, stable disease (32/77; 41.6%) and disease progression (25/77; 32.5%) were more frequently observed than partial (18/77; 23.4%) and complete (2/77; 2.6%) response (Table 3). This finding is to be expected because systemic therapies rarely result in complete tumor response in patients with liver metastases from breast cancer [35]. Most of these patients have a poor prognosis, and systemic treatment is usually not curative but may prolong survival [30].
|
The results of our study might be useful in clinical practice and have an impact on patient treatment. The response of metastatic liver lesions to systemic treatment is an important indicator to either maintain or change the chemotherapeutic regimen. Generally, when imaging studies show complete response, partial response, or stable disease, the chemotherapy regimen is continued without replacing any treatment agents. Conversely, if imaging studies indicate disease progression, a change in the therapy is often initiated [6]. Until now, tumor size and the number of metastases were the only two generally accepted factors defining the response to systemic treatment. Our findings provide suggestive evidence that tumor vascularity as characterized on MRI studies should be a new factor in the analysis of tumor response in breast cancer patients with liver metastases, because a strong correlation between tumor hypervascularity and disease progression in the follow-up examinations was observed.
We acknowledge the limitations of our retrospective investigation. First and most important, this was a single-institution study and the sample size was small, which is reflected in the wide 95% CI for the odds ratio of tumor progression for hypervascular versus hypovascular lesions (odds ratio, 20.5; 95% CI, 5.1–83.5). Nonetheless, even the low, conservative end of the confidence interval suggests a 5.1 times increased risk of disease progression in patients with hypervascular liver metastases. Second, we lack the correlation between histopathologic assessment of vascularity and MRI. Third, the time interval between follow-up MRI examinations was not standardized but at the discretion of the attending oncologist. Thus it is possible that patients with hypervascular metastases were more likely to receive follow-up examinations. However, the statistically significant association between hypervascularity of liver metastases and disease progression persisted in sensitivity analyses in which patients with the most MRI examinations were systematically removed, which indicates that clustering of MRI examinations in a patient does not threaten the validity of our findings. Another limitation was that the same radiologists viewed all MRI studies of the same patient at one setting. We thought that negligible expectation observation error occurred because the reviewers were blinded to the clinical picture, and we believed that in this initial study evaluating enhancement the reviewers should be aware of all the subtleties of disease change by having all studies available. Also, to minimize expectation bias, the investigators were unaware of the clinical course and laboratory parameters of the patients. Finally, disease progression in our study was assessed only for the liver. Systematic evaluation of other organ systems was not performed.
In summary, the results of this retrospective analysis provide suggestive evidence that hypervascularity of liver metastases as characterized on MRI predicts disease progression in breast cancer patients undergoing systemic therapy. MRI examinations may provide useful information for predicting tumor response to systemic treatment. The results of our pilot study must be confirmed by further prospective investigations with larger sample sizes and standardized treatment and follow-up schedules to determine whether the presence of hypervascular metastases on MRI might have therapeutic implications.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  J. P.B. O'Connor, G. C. Jayson, A. Jackson, D. Ghiorghiu, B. M. Carrington, C. J. Rose, S. J. Mills, R. Swindell, C. Roberts, C. L. Mitchell, et al. Enhancing Fraction Predicts Clinical Outcome following First-Line Chemotherapy in Patients with Epithelial Ovarian Carcinoma Clin. Cancer Res., October 15, 2007; 13(20): 6130 - 6135. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. H. Khandani and R. C. Semelka Can the Relation between Gadopentetate Dimeglumine and FDG Uptake in Colorectal Liver Metastases Be Used Clinically? Radiology, October 1, 2005; 237(1): 1 - 2. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
