A comparison of effectiveness of the contrast enhanced computed tomography with magnetic resonance imaging in the differential diagnosis of clear cell renal carcinoma

This retrospective, observational, single-center study was performed in the Department of Diagnostic Imaging in the University Hospital of Krakow and included 51 patients with indeterminate SRMs detected on CECT imaging. Final diagnoses were made on histopathologic results as standard reference (n = 38) or based on clinical and imaging features (diagnosis of lipid-poor angiomyolipoma or non-ccRCC masses without exact subtype information) if a regression or complete lack of progression was observed during follow-up (minimum 36 months). Forty-two of these patients also had available MRI results. The study was approved by the Ethics Committee and the requirement for patient informed consent was waived (OIL/KBL/51/2023).

All CECT examinations were performed using multirow helical scanners (64, 128, or 256 rows) and the following parameters: slice thickness 1.25–2.5 mm, tube voltage 120 kV, tube current-time product 50–250 mAs, matrix 512 × 512 pixels. Computed tomography images in non-enhanced, corticomedullary, nephrographic, and excretion phases were obtained. Non-ionic agents were injected at a rate of 3.5–4.0 ml/s with volume 450 mg I/kg of body weight. In solid lesions, the region of interest for attenuation measurement was positioned over the area that exhibited the greatest postcontrast enhancement. Computed tomography studies were assessed by two radiologists with at least 3 years of experience, blinded to the final diagnosis.

The assessed CECT features were categorized into four groups: (1) Anatomical, including the mass’s location in the right kidney or left kidney, determined by its position within the renal parenchyma; (2) Morphological, encompassing bean-like shape (an elongated, irregular contour resembling a bean), ball-like shape (a rounded, spherical appearance), and heterogeneous mass (non-uniform density indicating varied tissue components such as necrosis or viable tumor), all assessed qualitatively; (3) Enhancement, comprising significant enhancement (> 15 Hounsfield Units [HU], measured as the difference between pre- and post-contrast phases to reflect vascularity), heterogeneous enhancement (non-uniform contrast uptake across the mass), washout pattern (rapid contrast clearance, defined as a decrease of at least 20 HU between corticomedullary and nephrographic phases), prolonged pattern (sustained or slow contrast enhancement into the delayed phase), enhancement in corticomedullary phase (contrast uptake 30–60 seconds post-injection), difference in enhancement between corticomedullary and native phase (increase in HU from pre-contrast to corticomedullary phase), difference in enhancement between corticomedullary and nephrographic phase (HU change from corticomedullary to nephrographic phase, typically 70–120 seconds post-injection), absolute washout (calculated as [(HU corticomedullary – HU nephrographic) / (HU corticomedullary – HU native)] × 100, reflecting proportional contrast clearance), relative washout (calculated as [(HU corticomedullary – HU nephrographic) / HU corticomedullary] × 100, indicating percentage clearance from peak enhancement), difference in enhancement between nephrographic and native phase (HU increase from pre-contrast to nephrographic phase), and enhancement in nephrographic phase (contrast uptake in the nephrographic phase), all evaluated visually and quantitatively using region-of-interest measurements; and (4) Secondary Features, including tumor-feeding vessels (visible vascular structures supplying the mass, indicating angiogenesis), necrosis area (nonenhancing, low-density regions consistent with tissue death), and calcification (high-density calcium deposits within the mass), identified through visual inspection and density analysis.

Magnetic resonance imaging studies were performed using 1.5T or 3.0T scanners and included all the necessary sequences for lesion assessment according to the ccLS (Clear Cell likelihood score) v2.0: axial and coronal 2D T2w single shot acquisitions, axial 2D T1w gradient echo in and out of phase for chemical shift imaging, pre- and (dynamic) postcontrast 3D T1w SPGR with fat saturation including delayed scans and diffusion-weighted imaging. Detailed information regarding our MRI protocol and analyzed MRI parameters was described previously.¹³ The MRI parameters assessed included: (1) Intense corticomedullary enhancement, defined as strong contrast uptake in the corticomedullary phase (typically 30–60 seconds post-gadolinium injection), reflecting robust vascularity; (2) Clear Cell Likelihood Score (ccLS) = 5, a qualitative score from a standardized system assessing the highest likelihood of ccRCC based on MRI features such as enhancement patterns, signal intensity, and heterogeneity, assigned by radiologists; and (3) ccLS = 4/5, a combined score reflecting high to very high likelihood of ccRCC, similarly derived from visual interpretation of MRI characteristics; (4) arterial-to-delayed ratio (ADER), obtained as the ratio of difference between signal intensity in the corticomedullary phase and pre-contrast images to the difference in signal intensity in delayed phase and pre-contrast images; and (5) T1 SI measured as the ratio of tumor SI to renal cortex SI. According to our previous results, we used the following cut-off point for the MRI parameters: ADER > 0.99 and T1 SI (signal intensity) ratio < 0.73.¹³

The Student’s t-test or the Mann-Whitney test was used to compare continuous variables depicted as means ± standard deviations or medians and interquartile ranges (IQR). Pearson’s χ2 test or Fisher’s exact test was used to evaluate categorical variables, which are presented as numbers and percentages. Receiver operating characteristic (ROC) analysis was performed to calculate the area under the curve (AUC) and was used to determine which of the selected CECT features are discriminators between ccRCC and other types of SRMs. A logistic regression was performed in order to determine which of the CECT and MRI parameters are significant predictors of ccRCC. Specificity, sensitivity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and negative likelihood ratio were calculated for each tested criterion. A p-value of 0.05 or less was considered statistically significant. Statistical analyses were performed with IBM SPSS Statistics (version 24, IBM Corp., Armonk, NY, USA). Confidence intervals (CI) were calculated using MEDCALC (free statistical calculators).

A total of 51 patients (49% male) with indeterminate solid SRMs were retrospectively enrolled in the study. In the majority of patients, symptoms indicating the presence of a renal mass were not present (Figure 2). Patients with ccRCC did not differ significantly from patients with other than ccRCC etiology of renal masses in terms of presented symptoms: macroscopic hematuria (2 [11.1%] vs. 2 [6.1%], p = 0.607), flank pain (2 [11.1%] vs. 5 [15.2%], p = 1), microscopic hematuria (1 [5.6%] vs. 2 [6.1%], p = 1).

Figure 2.

Prevalence of observed symptoms in patients with ccRCC and non-ccRCC.

ccRCC = clear cell renal cell carcinoma; non-ccRCC = other than clear cell renal cell carcinoma

The population was also examined for the presence of individual risk factors. Different risk factors were more commonly detected in the ccRCC group when compared to the non-ccRCC group (94.4% vs. 66.7%, p = 0.037). In the ccRCC group, there were significantly more smoking patients than in the non-RCC group (11 [61.1%] vs. 9 [27.3%], p = 0.018). Similarly, there were more hypertensive patients in the ccRCC group when compared to the non-ccRCC group (14 [77.8%] vs. 16 [48.5%], p = 0.042) (Table 1).

Many of the compared CECT parameters differed significantly between the ccRCC group and the non-ccRCC group, as detailed in Table 2. Heterogenous mass (indicating non-uniformity in density due to varied tissue components such as necrosis or viable tumor) was observed significantly more often in the ccRCC group (88.9% vs. 60.6%, p = 0.034) compared to the non-ccRCC group, where homogenous mass dominated (11.1% vs. 39.4%, p = 0.034), suggesting greater structural complexity in ccRCC. Similarly, heterogeneous enhancement (where contrast uptake varies across the mass due to differing vascularity or tissue composition) was more frequent in the ccRCC group than the non-ccRCC group (83.3% vs. 51.5%, p = 0.025) where homogenous enhancement dominated (16.7% vs. 48.5%, p = 0.025), reflecting varied contrast uptake within ccRCC lesions. Moreover, presence of tumor-feeding vessels, identified as distinct vascular structures supplying the mass and reflecting increased angiogenesis, was also significantly greater in the ccRCC group (33.3% vs. 3.0%, p = 0.006), highlighting increased angiogenesis in ccRCC. Presence of a washout pattern, defined as a rapid decrease in enhancement (at least 20 HU difference between corticomedullary and nephrographic phases), indicative of quick contrast clearance, was more commonly observed in patients with ccRCC when compared to non-ccRCC etiology of SRMs (83.3% vs. 33.3%, p < 0.001). On the other hand, prolonged enhancement, characterized by sustained or slow contrast enhancement into the delayed phase, suggesting less rapid clearance, was detected more commonly in the non-ccRCC group when compared to the ccRCC group (66.7% vs. 16.7%), p < 0.001).

The anatomical location of the mass in either the right or left kidney did not significantly differ between the ccRCC and non-ccRCC groups. Morphologically, masses were classified as either bean-like, with an elongated, irregular contour, or ball-like, with a rounded, spherical appearance, but these shape differences between the ccRCC and non-ccRCC groups were not significant (p = 0.393), suggesting morphology alone is not discriminative. Similarly, significant enhancement, defined as an increase in density greater than 15 Hounsfield Units (HU) post-contrast, reflecting substantial vascularity was not significant.

There were also no significant differences between the ccRCC and non-ccRCC groups in necrosis area (defined as non-enhancing, low-density regions within the mass consistent with tissue death), calcification (identified as high-density calcium deposits within the mass), thrombosis (characterized as a thrombus within associated vasculature, e.g., renal vein, appearing as a filling defect), or macroscopic fat (described as very low-density fat tissue within the mass), indicating these features are not distinguishing between the two etiologies.

Receiver operating curve analysis revealed that from the selected CECT features reflecting contrast enhancement, relative washout (difference in enhancement between corticomedullary and nephrographic phase / enhancement in corticomedullary phase)*100 had the highest AUC value (0.850) for the prediction of ccRCC occurrence (Table 3). Receiver operating curves are depicted on Figure 3.

Figure 3.

ROC analysis of the selected CT parameters.

Univariate logistic regression analysis demonstrated that smoking and different CECT parameters were predictors of ccRCC (Table 4). In multivariate analysis, only relative washout and smoking remained as significant predictors of ccRCC occurrence (OR: 1.19, 95% CI: 1.01–1.41, p = 0.042; OR: 7.50, 95% CI: 1.13–49.88, p = 0.04; Table 4).

Using the Youden index, we determined the proposed cut-off point for relative washout in the ROC analysis to be 11.54 (Figure 4). Then, we compared the relative washout parameter with the newly established cut-off value with selected novel MRI score/parameters (Table 5). In results we found that only T1 SI ratio < 0.73 (defined as the lesion’s signal intensity relative to the renal cortex on precontrast T1-weighted MRI), ccLS = 5 (a qualitative score indicating the highest likelihood of ccRCC based on MRI features like enhancement and heterogeneity), and relative washout > 11.5 (calculated as the percentage decrease in HU from peak to delayed phase on CECT reflecting rapid contrast clearance) were different in patients with ccRCC when compared to the non-ccRCC group (27.8% vs. 3.0%, p = 0.007; 16.7% vs. 0.0%, p = 0.024; 88.9% vs. 30.3%, p < 0.001; Table 5). The parameters that did not significantly differ between the ccRCC and non-ccRCC groups included intense corticomedullary enhancement (defined as strong contrast uptake in the corticomedullary phase of MRI), the ADER > 0.99 (a diffusion-weighted imaging metric comparing Apparent Diffusion Coefficient before and after contrast), and ccLS = 4/5 (a combined score reflecting high to very high likelihood of ccRCC).

Figure 4.

The cut-off point is determined in the ROC analysis using the Youden index.

Youden index = 0.59; Suggested cut-off: 11.54

In the next step, we put these MRI variables, relative washout with the new cut-off point, and smoking into multivariable analysis. We observed that only T1 < 0.73 and relative washout >11.5 were independent predictors of ccRCC occurrence (OR: 30.86, 95% CI: 1.58–600.26, p = 0.024; OR: 15.36, 95% CI: 1.52–155.16, p = 0.021). When we compared CECT and MRI parameters, we noticed that the best accuracy (76.5%) and sensitivity (88.9%) were observed for relative washout, with a good specificity of 69.7%. Table 6 shows specificity, sensitivity, accuracy, positive likelihood ratio, and positive predictive value for these selected imaging parameters.