Evaluation of central non-enhancement in solid renal masses

Author

Lu Mao

Published

February 13, 2025

Statistical analysis

Categorical variables are summarized by frequency and percentage; continuous variables are summarized by median and interquartile range (IQR). The agreement between two readers is assessed using the kappa coefficient for binary variables and the intraclass correlation coefficient (ICC) for continuous variables. The Wilcoxon rank sum test is used to compare histologic percent necrosis between groups with and without central non-enhancement (CNE). The correlation between volumetric non-enhancement to mass ratio and histologic percent necrosis is assessed using the Spearman correlation coefficient. Associations of histologic percent necrosis or volumetric non-enhancement to mass ratio with clinical outcomes are evaluated using the Spearman correlatione or Wilcoxon rank sum test as appropriate. P-values less than 0.05 are considered statistically significant. All statistical analyses are performed using R version 4.1.1 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Patient characteristics

Patient characteristics are summarized by site and overall in Table 1 below.

Table 1: Patient characteristics
Characteristic Overall, N = 771 UTSW, N = 511 MGB, N = 261
Age (years) 66 (58, 73) 67 (58, 74) 62 (55, 69)
Gender


    Female 19 (25%) 13 (25%) 6 (23%)
    Male 58 (75%) 38 (75%) 20 (77%)
Clinical T Stage


    T1b 32 (63%) 32 (63%) 0 (NA%)
    T2 19 (37%) 19 (37%) 0 (NA%)
    Unknown 26 0 26
WHO/ISUP Grade


    Grade 1 3 (3.9%) 1 (2.0%) 2 (7.7%)
    Grade 2 29 (38%) 22 (43%) 7 (27%)
    Grade 3 23 (30%) 14 (27%) 9 (35%)
    Grade 4 13 (17%) 7 (14%) 6 (23%)
    Not applicable 9 (12%) 7 (14%) 2 (7.7%)
AJCC TNM Stage


    Stage I 33 (43%) 24 (47%) 9 (35%)
    Stage II 12 (16%) 6 (12%) 6 (23%)
    Stage III 27 (35%) 19 (37%) 8 (31%)
    Stage IV 5 (6.5%) 2 (3.9%) 3 (12%)
Pathology Report Necrosis > 0% 38 (49%) 13 (25%) 25 (96%)
Rhabdoid Features


    Present 7 (9.1%) 5 (9.8%) 2 (7.7%)
    Not Present 66 (86%) 42 (82%) 24 (92%)
    Unknown 4 (5.2%) 4 (7.8%) 0 (0%)
Sarcomatoid Features


    Present 4 (5.2%) 3 (5.9%) 1 (3.8%)
    Not Present 71 (92%) 46 (90%) 25 (96%)
    Unknown 2 (2.6%) 2 (3.9%) 0 (0%)
Presence of CNE (Reader 1) 64 (83%) 38 (75%) 26 (100%)
Presence of CNE (Reader 2) 70 (91%) 44 (86%) 26 (100%)
1 Median (IQR); n (%)

Reader agreement

For the presence of central non-enhancement, the two reader assessments are cross-tabulated in Table 2 below. The kappa coefficient is 0.546 (p<0.001), indicating a moderate agreement.

Table 2: Cross-tabulation of central non-enhancement presence by reader
Presence of CNE (Reader 2) Total p-value1
No Yes
Presence of CNE (Reader 1)


<0.001
    No 6 (7.8%) 7 (9.1%) 13 (17%)
    Yes 1 (1.3%) 63 (82%) 64 (83%)
Total 7 (9.1%) 70 (91%) 77 (100%)
1 Fisher’s exact test

For the quantitative volumetricnon-enhancement to mass ratio, the scatter plot of the two reader assessments is shown in Figure 1 below. The ICC indicates a stronger agreement.

Figure 1: Scatter plot of non-enhancement to mass ratio by reader

CNE vs. histologic percent necrosis

The following operations are applied to non-standard values of Pathology Report Percent Necrosis:

  • <20 \(\to\) 20;
  • 15-20 \(\to\) 17.5.

Binary presence/absence of CNE

Consider Reader 1, Reader 2, and combinations of their assessments by AND and OR. The histologic percent necrosis is compared between groups with and without CNE for each definition of CNE presence. The results are summarized in Table 3 and Figure 2 below. Most of histologic percent necrosis is zero in the absence of CNE, and all comparisons are statistically significant.

Table 3: Wilcoxon rank sum test of histologic necrosis (%) vs. presence of CNE (N; median (IQR))
No Yes P
Reader 1 N = 13; 0 (0, 0) N = 64; 5 (0, 15.6) 0.005
Reader 2 N = 7; 0 (0, 0) N = 70; 5 (0, 15) 0.012
Both present N = 14; 0 (0, 0) N = 63; 5 (0, 16.2) 0.003
Either present N = 6; 0 (0, 0) N = 71; 5 (0, 15) 0.021
Figure 2: Boxplot of histologic necrosis by reader-rated presence of CNE

Volumetric non-enhancement to mass ratio

The volumetric non-enhancement to mass ratio, by each reader and their average, is correlated with histologic percent necrosis in Figure 3. The correlations are moderately strong and statistically significant.

Figure 3: Correlation of non-enhancement to mass ratio with histologic necrosis (%)

Histologic percent necrosis vs. clinical outcomes

The histologic percent necrosis is compared between groups stratified by WHO/ISUP grade, AJCC TNM stage, rhabdoid features, and sarcomatoid features. The results are summarized in Table 4 (binary \(0\%\) vs. \(>0\%\)) and Figure 4 (quantitative) below.

Table 4: Association of histologic necrosis (%) vs. clinical outcomes
Characteristic 0%, N = 391 >0%, N = 381 p-value2
WHO/ISUP Grade

<0.001
    Grade 1 1 (2.9%) 2 (6.1%)
    Grade 2 21 (60%) 8 (24%)
    Grade 3 12 (34%) 11 (33%)
    Grade 4 1 (2.9%) 12 (36%)
    Unknown 4 5
AJCC TNM Stage

0.035
    Stage I 22 (56%) 11 (29%)
    Stage II 3 (7.7%) 9 (24%)
    Stage III 13 (33%) 14 (37%)
    Stage IV 1 (2.6%) 4 (11%)
Rhabdoid Features

0.11
    Not Present 35 (97%) 31 (84%)
    Present 1 (2.8%) 6 (16%)
    Unknown 3 1
Sarcomatoid Features

0.054
    Not Present 38 (100%) 33 (89%)
    Present 0 (0%) 4 (11%)
    Unknown 1 1
1 n (%)
2 Fisher’s exact test
Figure 4: Boxplot of histologic necrosis by clinical outcomes
  • Due to large noise in \(>0\%\) group, quantitative test generally less significant than binary test in Table 4.

CNE vs. clinical outcomes

The reader-averaged non-enhancement to mass ratio is compared between groups stratified by WHO/ISUP grade, AJCC TNM stage, rhabdoid features, and sarcomatoid features. The results are summarized in Figure 5 and Table 5 below.

Figure 5: Correlation of non-enhancement to mass ratio with clinical outcomes
Table 5: Association of non-enhancement to mass ratio (reader average) vs. clinical outcomes
Reader-Averaged Ratio by WHO ISUP Grade
Characteristic Grade 1, N = 31 Grade 2, N = 291 Grade 3, N = 231 Grade 4, N = 131
Median (IQR) 0.03 (0.02, 0.04) 0.01 (0.01, 0.03) 0.05 (0.01, 0.14) 0.07 (0.01, 0.15)
1 Median (IQR)
Reader-Averaged Ratio by AJCC TNM Stage
Characteristic **Stage I **, N = 331 **Stage II **, N = 121 **Stage III **, N = 271 **Stage IV **, N = 51
Median (IQR) 0.01 (0.00, 0.05) 0.04 (0.01, 0.18) 0.02 (0.01, 0.12) 0.16 (0.08, 0.20)
1 Median (IQR)
Reader-Averaged Ratio by Rhabdoid Features
Characteristic Not Present, N = 661 Present, N = 71
Median (IQR) 0.02 (0.01, 0.13) 0.07 (0.01, 0.13)
1 Median (IQR)
Reader-Averaged Ratio by Sarcomatoid Features
Characteristic Not Present, N = 711 Present, N = 41
Median (IQR) 0.02 (0.01, 0.12) 0.07 (0.01, 0.15)
1 Median (IQR)
  • Moderate and (borderline) significant correlations with WHO/ISUP grade and AJCC TNM stage;
  • Weak to no correlation with rhabdoid and sarcomatoid features.