Low-dose imaging technique (LITE) MRI: initial experience in breast imaging

Abstract

Objectives:

To compare a low-dose dynamic contrast-enhanced breast MRI protocol (LITE MRI) to standard-dosage using a dual-dose injection technique.

Methods:

8 females with a total of 10 lesions with imaging features compatible with fibroadenoma were imaged using a dual-dose dynamic contrast-enhanced-MRI (DCE-MRI) technique. After pre-contrast scans, 15% of a standard dose of contrast was administered; approximately 10 min later, the remaining 85% of the standard dose was administered. Enhancement kinetic parameters, conspicuity and signal-to-noise ratio were measured quantitatively.

Results:

One lesion showed no enhancement in either DCE series. All nine of the enhancing lesions were visualized in both the low-dose and standard-dose images. While the (low-to-standard) ratio of contrast doses was roughly 0.18, this did not match the ratios of kinetic parameters. Lesion conspicuity and enhancement rate were both higher in the low-dose images, with (low-to-standard) ratios 1.5 ± 0.1 and 1.2 ± 0.4, respectively. The upper limit of enhancement (ratio 0.3 ± 0.1) and signal-to-noise ratio (ratio 0.5 ± 0.1) were higher in the standard-dose images, but less than expected based on the ratio of the doses.

Conclusions:

This preliminary study demonstrates that LITE MRI has the potential to match standard DCE-MRI in the detection of enhancing lesions. Additionally, LITE MRI may enhance sensitivity to contrast media dynamics.

Advances in knowledge:

Lower doses of MRI contrast media may be equally effective in the detection of breast lesions, and increase sensitivity to contrast media dynamics. LITE MRI may help increase screening compliance and long-term patient safety.

Introduction

Dynamic contrast-enhanced MRI (DCE-MRI) is highly sensitive to breast cancer (93.2% pooled sensitivity¹). During a standard DCE-MRI acquisition, a series of T ₁ weighted images are acquired before and after the intravenous administration of a gadolinium (Gd) based contrast agent (GBCA). Gadolinium distributes in the extracellular space and accumulates in tissues with rich vascularity, high vascular permeability and/or an expanded interstitial space.² This mechanism of distribution allows for the differentiation between benign and malignant lesions based on kinetics and degree of enhancement.^3,4

Recently, it was discovered that gadolinium can deposit in the brain and other body parts in patients with normal renal function.^5–7 Although no long-term consequences for tissue toxicity, patient health, or clinical symptomatology have been identified to date, this is of growing concern in the patient and medical community. The US Food and Drug Administration issued a new drug safety warning of the deposition of gadolinium, directing healthcare providers to consider the retention properties of GBCAs.⁸ The European Medicines Agency has suspended the authorization of all but one linear GBCA except in certain specialized applications.⁹ This accumulation is dose-dependent with higher deposition of Gd seen in patients who receive greater cumulative doses of the contrast agent. Reducing the amount of contrast administered per imaging session could help alleviate some of these concerns.

The current standard dose of GBCAs administered is 0.1 mmol/kg. This dose was determined in the 1980s after initial pre-clinical and clinical experience showed this dose to be well-tolerated and effective for imaging—but this early work did not determine the minimum effective dose.¹⁰ The field strength of scanners used for clinical scans at the time (0.35 – 0.5 T) was markedly lower than modern magnets (1.5 and 3 T). The native T ₁ of tissues increases with increasing field strength; leading to higher signal enhancement in T ₁ weighted images in the presence of contrast. Other improvements in MRI technology also mean that more modest enhancement is better visualized in modern scanners. Thus the standard dose of 0.1 mmol/kg may not be optimal for modern-day breast DCE-MRI. Low doses of contrast agent also result in reduced water exchange and T ₂* effects, facilitating measurements of the arterial input function (AIF) and quantitative analysis.¹¹ The AIF is a measure of the concentration of contrast media in the arterial blood supply. Because concentrations in the arteries at standard doses of contrast media are relatively high, it is difficult to obtain accurate estimates of the AIF near peak concentration. Obtaining an accurate AIF is a critical step in the pharmacokinetic analysis of lesions, which leads to parameters descriptive of tumor physiology that may aid in diagnosis and prediction of response to therapy.¹²

To address these issues, we propose a low-dose imaging technique (LITE) for breast DCE-MRI. In this manuscript, we describe our initial experience with LITE MRI using a 0.015 mM/kg dose of GBCA, and compare it to images acquired with standard doses of contrast media. Here we report on LITE MRI of fibroadenomas. Fibroadenomas have enhancement kinetics similar to those of cancers (i.e. rapid initial enhancement), but are relatively homogeneous. Therefore, they provide a useful starting point for evaluating the LITE protocol that avoids confounding factors associated with the inter- and intralesion heterogeneity of cancers.

Methods and materials

8 patients (ages 18–60 years) with a total of 10 lesions with imaging features most compatible with a fibroadenoma were imaged under an IRB-approved protocol. The lesions ranged in size (as measured on diagnostic ultrasound) from 0.5 to 2 cm. All lesions were either biopsy-proven or clinically confirmed to be benign over a 2-year time period through prior imaging. All images were acquired using a 3 T Ingenia scanner (Philips Healthcare, Netherlands) and a dedicated 16-channel breast coil (Invivo, Gainesville, Florida). Females were not scanned at a specific time in their menstrual cycle. Imaging consisted of non-contrast sequences (including T ₂ weighted images and a variable flip angle T ₁-mapping sequence), and DCE-MRI. During the DCE series, two types of images were acquired: standard clinical images (high spatial resolution and low temporal resolution of 61–79 s), and "ultrafast" images¹³ (lower spatial resolution and high temporal resolution of 3.2–3.6 s). Imaging parameters for these sequences are shown in Table 1. To evaluate low-dose images and compare them to imaging with a standard dose of contrast media, the contrast was delivered in two sequential administrations. First, 15% of the standard dose of contrast media—0.1 mm/kg of gadobenate dimeglumine (MultiHance, Bracco, Italy)—was administered followed by the acquisition of a series of DCE images (Figure 1) consisting of the standard clinical images and the "ultrafast" images. Approximately, 10 min following the first injection, the remaining 85% of the contrast dose was administered followed by the same imaging protocol as the one subsequent to the first contrast administration. GBCA was delivered using a power injector with a flow rate of 2 ml s⁻¹, followed by 20 ml of saline flush at 2 ml s⁻¹. For the purposes of this manuscript, 85% of the recommended standard dose of contrast media will be referred to as a "standard" dose.

Table 1. .

Imaging parameters for the DCE protocols used

Imagingparameter	Ultrafast	High spatial resolution
TR/TE (ms)	2.8/1.4	4.7/2.4
Acquisition voxel size (mm3)	1.5 × 1.5×4.0	0.8 × 0.8×1.6
SENSE acceleration factor (RL)	3	2.5
SENSE acceleration factor (FH)	2	2
Partial Fourier factor	0.65 (ky); 0.7 (kz)	0.85 (ky); 1 (kz)
Temporal resolution range (s)	3.2–3.6	61–79.5
Number of slices	88–110	220–275
Flip angle	10⁰
Field of view (mm)	320–360
Fat suppression method	SPAIR

DCE, dynamic contrast enhanced; TE, echo time; TR, repetition time.

Figure 1. .

Diagram of the protocol used for the DCE-MRI acquisition. The first injection consisted of 15% of the standard dose (0.1 mM/kg), the entire protocol shown above is then repeated with the remaining 85% of the standard dose of contrast media, approximately 10 min after the first injection. DCE, dynamic contrast-enhanced.

Semi-quantitative kinetic analysis was performed by fitting the mean relative signal enhancement in each lesion to an exponential empirical mathematical model,^13,14

$$RE\left(t\right)=A\left(1-\exp \left(-\alpha \left(t-t_0\right)\right)\right);fort>t_0$$ (1)

where A is the upper limit of the relative enhancement (RE), α (s⁻¹) is the relative enhancement rate (i.e., a measure of the speed of signal increase), and t₀ is the time-to-enhancement. From these parameters, the initial area under the uptake curve is obtained by integrating equation (1) out to desired time (in this case 30 s). Average enhancement values for each lesion were obtained by manually placing regions of interest (ROIs) around the entire enhancing lesion in all slices where the lesion was visible. For the second injection, the relative enhancement was calculated relative to the delayed phase of the first injection, in order to minimize the effect of residual contrast that may have been present in the lesion at the time of the second injection.

The conspicuity of each lesion on the DCE images was evaluated quantitatively by taking the ratio of the signal increase in the lesion to the signal increase in the surrounding uninvolved parenchyma. Mean conspicuity was then calculated by averaging over all time-points after lesion enhancement began.

The signal-to-noise ratio (SNR) of enhancement was calculated for each lesion for both sets of images. SNR was defined as the post-contrast signal in each voxel divided by the standard deviation of the pre-contrast signal as measured in the series of baseline images. Mean SNR values were calculated by averaging over all time-points after the lesions began to enhance.

All analysis was performed with in-house built software implemented in MATLAB (The Mathworks, Natick, MA); clinical evaluation of the lesions was performed on the same workstations used for routine clinical scans. The Wilcoxon signed-rank test was used to assess statistical significance of differences in parameters measured from both DCE-MRI series in each subject.

Results

9 of the 10 lesions had measurable enhancement on DCE-MRI. One lesion had no measurable enhancement either in the low- or standard-dose images. Motion artifacts affected the semi-quantitative kinetic analysis of a lesion in another case and it was excluded. Representative images for both low and standard dosage are shown in Figure 2. While the absolute enhancement was higher in the standard-dose images, the lesions are well visualized in the low-dose images. Due to the lower background parenchymal enhancement (BPE) in the low-dose images, the conspicuity of the lesions is greater (Figure 3), as reflected in the quantitative measurement of lesion conspicuity. The mean conspicuity in the low-dose images was 1.48 ± 0.15 times higher than in the standard-dose images (p < 0.05).

Figure 2. .

Maximum intensity projections of subtractions from ultrafast DCE depicting initial enhancement for both low and standard dose images. Times are expressed relative to each contrast media administration. DCE, dynamic contrast enhanced.

Figure 3. .

Signal increase (post-contrast minus pre-contrast in the lesion and surrounding uninvolved parenchyma. (a, b) Low dose subtraction images acquired at roughly 2 and 4.5 min after contrast injection, respectively. (d, e) Standard dose subtraction images acquired at the times listed above. (c, f) Signal increase time-courses for the lesion (solid line) and background parenchyma (dashed line), vertical dotted lines indicate the times at which the images shown were acquired.

Boxplots of the ratios of the parameters measured are shown in Figure 4. While the ratio of the contrast doses administered was roughly 0.18 (low-to-standard), this was not reflected in the ratios of the enhancement parameters, which were larger (more favorable) than the dose ratio (Table 2). In fact, the average enhancement rate measured from the low-dose images was higher than that of the standard dose, but this difference was not significant (p = 0.15). Lesion time-to-enhancement was similar for both doses, without a significant difference (p = 0.38). As expected, the average post-contrast SNR was higher in the standard-dose images (p = 0.008). However, the difference was smaller than what could be expected based on the ratio of the doses of contrast. Plots of representative lesion enhancement curves are shown in Figure 5a–b.

Figure 4. .

Boxplots of the ratios of the parameters measured for each lesion (median value shown in each box). The horizontal dashed line indicates the ratios of contrast doses administered. Low-to-standard dose ratio values are shown on the left y-axis, and corresponding standard-to-low values on the right y-axis.

Table 2. .

Summary of parameters measured from LITE and standard dose images (values expressed as mean ± standard deviation)

Parameter	Low Dose	Standard Dose	Ratio (Low-to-standard)
Contrast media dose (mM/kg)	0.015	0.085	0.18
Upper limit of relative enhancement (%) a	34.6 ± 6.8	115.2 ± 22.1	0.31 ± 0.07
Enhancement rate (%/s)	3.4 ± 2.6	2.6 ± 1.6	1.22 ± 0.41
Time-to-enhancement (s)	35.2 ± 8.1	37.2 ± 5.5	0.94 ± 0.20
Initial area under the curve (integrated to 30 s) (% s) a	3.55 ± 2.28	10.55 ± 6.41	0.33 ± 0.07
Lesion conspicuitya	7.72 ± 5.27	5.47 ± 4.56	1.48 ± 0.15
Mean post-contrast lesion SNR a	179.6 ± 165.6	377.8 ± 381.7	0.53 ± 0.13

LITE, low-dose imaging technique; SNR, signal-to-noise ratio.

^a p < 0.05

Figure 5. .

Example signal time courses for low and standard doses of contrast media: (a) signal throughout the entire DCE series for one lesion (dotted lines indicate injections of contrast media); (b) relative signal enhancement normalized to time of each injection for the lesion shown in (a); (c) signal intensity measured in the aorta of a representative case.DCE, dynamic contrast-enhanced.

Semi-quantitative AIFs (i.e. in units of signal intensity) are shown in Figure 5c. AIFs extracted from the low-dose cases are sharper and correspond better to the expected shape,¹⁵ with the first and second passes identifiable. In the standard-dose images the actual shape of the AIF is not captured.

Discussion

In this proof-of-concept study, we found that low-dose breast imaging is clinically feasible and may offer some advantages over standard-dose imaging. All eight of the enhancing lesions were visible on both the low-dose and standard-dose images. Ultrafast DCE-MRI increases the diagnostic utility of LITE by acquiring images at early times after the low-dose contrast media injection, when the conspicuity of lesions is greater. Ultrafast imaging also allows for the accurate measurement of early enhancement kinetics. While ultrafast protocols are not widely used in routine clinical practice today, there is increasing interest in the advantages they offer, for example greater lesion conspicuity in early post-contrast images. This has led to some institutions—ours included—to adopt ultrafast imaging during the initial phase of routine clinical DCE-MRI acquisitions. Quantitative analysis of the lesions demonstrate that the lesion conspicuity and average signal enhancement rate were more favorable in the low-dose imaging.

All lesions imaged in this preliminary study were either biopsy-proven fibroadenomas or clinically confirmed benign lesions with features most compatible with a fibroadenoma. This allowed us study a uniform population of lesions with enhancement kinetics similar to cancer, so that the data that could be interpreted without confounding factors, such as non-uniformity and intralesional inhomogeneity typically seen in malignant lesions. Future studies will be expanded to include malignant lesions, to determine whether low-dose can be diagnostically equivalent in both the pre-invasive and invasive cancer setting.

Breast MRI is highly sensitive for the detection of breast cancer. However, in the setting of moderate or marked BPE, it is possible that a small enhancing mass or area of non-mass-enhancement may be masked. Previous studies that reported undetected cancerous lesions on MRI indicated that BPE was a factor in false-negative interpretation.¹⁶ In this study, all lesions had greater conspicuity in low-dose imaging, due to preferential uptake of contrast within the lesion compared to the background parenchyma. This can be diagnostically advantageous in the clinical interpretation of the study.

Despite the standard-to-low ratio of doses being approximately 5.7, the upper limit of enhancement in the standard-dose images was roughly only three times higher than in the LITE series. While measurement of contrast media concentration could in part correct for these differences, the additional calculations required would increase error and noise. In addition, the effect from water exchange between the intracellular and extracellular spaces, and effects of contrast media on T ₂* would change the enhancement kinetics at higher concentrations of contrast media.¹⁷ In addition, the change in T ₂ due to the first 15% of the dose reduces the effect of the subsequent 85% of the contrast dose. Thus, the first 15% of the dose may provide much greater sensitivity to contrast media dynamics than the following 85%. The results demonstrate that low doses are feasible for lesion detection and characterization.

The main limitation of this study is the low number of cases. We plan to validate our results in a larger cohort, including malignant lesions. This will include less vascular lesions, such as ductal carcinoma in situ. If the trends observed in the current study hold true, we believe that the lower signal in the background parenchyma in LITE MRI will translate to better visualization of malignant lesions—including less vascularized ones—than on routine MRI. Additionally, the enhancement rates measured on LITE were higher than with the larger dose of contrast media, this may also aid in differentiating ductal carcinoma in situ from parenchymal enhancement. The scans were performed on a 3 T scanner, whereas most clinical breast MRI’s are performed at 1.5 T. It is possible that the optimal dose may be greater for scans performed at lower magnetic field strengths. Future work will include studying different doses to optimize the dose for LITE MRI under different conditions, and will include native T ₁ and T ₂* measurements in order to estimate the concentration of contrast media in lesions.

In conclusion, the results of this pilot study show that LITE MRI has the potential to be equivalent to standard DCE-MRI in the detection of enhancing breast lesions. LITE MRI may provide enhanced sensitivity to contrast media dynamics. Growing concern related to gadolinium deposition may be addressed with low-dose contrast administration, increasing screening compliance and long-term patient safety.