4.1. ADC and ADC ratio values
Advances in diffusion MRI techniques have made this imaging modality very useful for examination of the liver. Echo-planar imaging techniques, parallel image acquisition, and the availability of multichannel coils played decisive roles in improving the quality of diffusion-weighted images by decreasing movement artifacts (attributable to heart rate, respiration, and intestinal motility), thus allowing interpretation of ADC maps, which was previously impossible.[24–26]
As evaluation of the hepatic parenchyma became technically possible, so characterization of liver lesions through DWI and ADC map became the object of a—relatively recent—growing body of research. Interest in the potential of this technique for characterization of focal liver lesions has grown as technology has advanced. With diffusion sequences becoming less and less susceptible to (especially movement) artifacts, scanners are now able to acquire images with higher b-values, maintaining a good signal-to-noise ratio and providing scans with higher resolution.
Our findings suggest that ADC and ADCr values can be used to distinguish benign solid liver lesions from malignant ones, with mean values of 1.42×10− 3 mm2/s and 1.13×10− 3 mm2/s in the benign and malignant groups, respectively (p < 0.001 and p < 0.05). ROC curve analysis yielded a cutoff point of 1.19×10− 3 mm2/s for ADC and 1.08×10− 3 mm2/s for ADCr. This finding is in line with the existing literature, in which ADC values have been reported as higher in benign lesions and lower in malignant lesions. [4, 5, 7, 8, 15, 27, 28] Our cutoff point, however, departs from the findings of some published studies. The search for optimal cutoff points to discriminate benign from malignant lesions has been the object of intense study in the literature, and still warrants further investigation. Indeed, different studies have reported variable values. This inconsistency in ADC cutoff points is probably attributable, among other reasons, to differences between scanners and differences in the parameters used to obtain diffusion-weighted images and ADC maps.[22, 29, 30]
In our subgroup analysis restricted to benign lesions, adenomata and FNH exhibited ADC values of 1.41×10− 3 mm2/s and 1.30×10− 3 mm2/s respectively; both are intermediate values, and are not significantly difference between one another or in comparison with the ADC values found in malignant lesions. The application of DWI and ADC mapping in the characterization of certain benign solid liver lesions and the potential of this technique to distinguish benign lesions from malignant ones is still a topic of discussion. Solid benign lesions, such as FNH and adenomata, may also exhibit diffusion restriction when compared to normal hepatic parenchyma, depending, among other reasons, on the proportion of their constituent tissues. The ADC values of these lesions, especially in the case of adenoma and FNH, overlap with those of malignant lesions.[1, 8]
Consequently, some authors advocate that diffusion techniques should not be used for discrimination of adenomata and FNH; in these cases, diffusion-weighted sequences would be evaluated in conjunction with other acquisition sequences within the standard liver MRI protocol. [31–33] However, the literature has demonstrated the role of ADC values for assessing the degree of differentiation and microvascular invasion in some types of cancer, with lower values in less differentiated tumors. [34–37] Investigators have searched for a similar phenomenon in benign lesions, in an attempt to demonstrate differences in ADC values among subtypes of these lesions, but such research is still incipient. We believe that additional, robust studies that provide for the classification of benign liver lesions within subtypes that respect their constitutional characteristics could help elucidate this matter.
In conclusion, the results of our study show diffusion-weighted imaging and ADC maps are useful and appropriate techniques for discrimination between benign solid lesions and malignant metastatic lesions, but not in isolation, at least not at the present time. Correlation is still required between the information provided by these sequences, the signal pattern of other MRI acquisitions, and the enhancement pattern of the observed lesions.
4.2. Analysis of sample characteristics
On comparative analysis, several patient characteristics were found to be associated with malignancy: advanced age, male sex, a known diagnosis of cancer, multiple lesions, larger lesions, and lower ADC and ADCr values.
After adjustment by the multivariate model, these correlations remained only for age (with a 2% increase in the odds of malignancy for each additional year of age) and known diagnosis of cancer (with a seven-fold increase in the odds of malignancy among patients with a history of cancer).
4.3. Study limitations
The core limitations of this study are those inherent to all single-center retrospective research. In addition, the limitation of the reference standard used for final diagnosis of benign lesions must be acknowledged. Despite good specificity and sensitivity of imaging, histology remains the gold standard. Nevertheless, we believe the absence of histological data in this type of research is justified, considering that liver biopsy is an invasive procedure of limited or no benefit in the evaluation of focal hepatic lesions deemed probably benign on imaging [3, 4, 7, 18, 20, 22, 23, 38].
The results of this study demonstrate the appropriateness of DWI and ADC mapping for characterization of benign solid lesions of the liver and to distinguish these masses from malignant metastatic lesions. The ADC and ADC ratio showed no statistical difference in discriminating benign focal lesions from liver metastases. These findings can be used to assist in differential diagnosis of focal liver lesions in the noncirrhotic patient.