Patient Characteristics
Out of 185 patients meeting the inclusion criteria, 152 patients (82%) had at least two CA-125 values available for calculation of their Std KELIM and were included in the study.
The clinical and biological characteristics of patients according to their platinum sensitivity status are presented in Table 1. Overall, 101 patients (66.4%) had a platinum-sensitive recurrence and 51 (33.6%) had a platinum-resistant recurrence. The median age was 63.1 years. 91.4% of the patients had serous adenocarcinoma, and all were high-grade epithelial tumors according to final histology. Optimal IDS, i.e. CC0 status, was achieved in 93 of 152 patients (62%).
Characteristics
|
Overall Pop
(N = 152)
|
Platinum Sensitive (N = 101)
|
Platinum Resistant (N = 51)
|
p value
|
Table 1
Patient characteristics based on platinum-sensitivity status
Age (in years)
|
63.1 (9.84)
|
62.8 (9.96)
|
63.6 (9.69)
|
0.64
|
Histology
|
|
0.92
|
Serous
|
139 (91.4%)
|
93 (92.1%)
|
46 (90.2%)
|
|
Clear cell
|
3 (2.0%)
|
2 (2.0%)
|
1 (2.0%)
|
|
Endometrioid
|
1 (0.7%)
|
1 (1.0%)
|
0 (0.0%)
|
|
Mucinous
|
2 (1.3%)
|
1 (1.0%)
|
1 (2.0%)
|
|
Undifferentiated
|
7 (4.6%)
|
4 (4.0%)
|
3 (5.9%)
|
|
FIGO stage
|
|
|
|
0.57
|
IIA
|
1 (0.7%)
|
1 (1.0%)
|
0 (0.0%)
|
|
IIB
|
2 (1.3%)
|
2 (2.0%)
|
0 (0.0%)
|
|
IIIA
|
2 (1.3%)
|
2 (2.0%)
|
0 (0.0%)
|
|
IIIB
|
2 (1.3%)
|
2 (2.0%)
|
0 (0.0%)
|
|
IIIC
|
99 (65.1%)
|
68 (67.3%)
|
31 (60.8%)
|
|
IVA
|
24 (15.8%)
|
13 (12.9%)
|
11 (21.6%)
|
|
IVB
|
22 (14.5%)
|
13 (12.9%)
|
9 (17.6%)
|
|
Type of NAC
|
|
|
|
0.40
|
Platinum and taxane
|
139 (91.4%)
|
94 (93.1%)
|
45 (88.2%)
|
|
Platinum only
|
1 (0.7%)
|
1 (1.0%)
|
0 (0.0%)
|
|
Other platinum-based NAC
|
12 (7.9%)
|
6 (5.9%)
|
6 (11.8%)
|
|
Number of Cycles of NAC
|
|
0.44
|
< 3
|
1 (0.7%)
|
1 (1.0%)
|
0 (0.0%)
|
|
3
|
4 (2.6%)
|
2 (2.0%)
|
2 (3.9%)
|
|
4
|
35 (23.0%)
|
27 (26.7%)
|
8 (15.7%)
|
|
5
|
10 (6.6%)
|
5 (5.0%)
|
5 (9.8%)
|
|
6
|
85 (55.9%)
|
56 (55.4%)
|
29 (56.9%)
|
|
> 6
|
17 (11.2%)
|
10 (9.9%)
|
7 (13.7%)
|
|
CA-125 at diagnosis
|
1757 (2375)
|
1677 (2434)
|
1914 (2271)
|
0.15
|
CA-125 before IDS
|
84 (264.7)
|
43 (69.93)
|
166 (440.6)
|
< 0.001
|
Std KELIM
|
0.79 (0.37)
|
0.86 (0.37)
|
0.66 (0.33)
|
< 0.001
|
Std KELIM*
|
|
|
|
0.004
|
Unfavorable
|
38 (25.0%)
|
19(18.8%)
|
19 (37.0%)
|
|
Intermediate
|
73 (48.0%)
|
47 (46.5%)
|
26 (51.0%)
|
|
Favorable
|
41 (27.0%)
|
35 (34.7%)
|
6 (12.0%)
|
|
RD after IDS
|
|
|
|
< 0.001
|
No CC0
|
57 (38.0%)
|
26 (26.0%)
|
31 (62.0%)
|
|
CC0
|
93 (62.0%)
|
74 (74.0%)
|
19 (38.0%)
|
|
Results are expressed in mean (standard deviation) or number of patients (percentage) |
CC0: absence of macroscopic postoperative residual disease; NAC: neoadjuvant chemotherapy; IDS: interval debulking surgery; Pop: population; RD: residual disease |
*: Thresholds used for these 3 categories were 0.5 and 1 (used in the study by You et al., corresponding to terciles of their population) |
Std KELIM
The mean Std KELIM was significantly lower in patients with platinum-resistant recurrence versus platinum-sensitive relapse patients at 0.66 vs 0.86 (p < 0.001) (Fig. 1). The mean Std KELIM was significantly lower in patients with platinum-resistant recurrence whether it was considered as a discrete covariate with the different thresholds (0.5, 0.6, 0.8 and 1) (Supplementary Data). The mean pre-IDS CA-125 level was 43 IU/mL in the platinum-sensitive relapse group and 166 IU/mL in the platinum-resistant recurrence group. In univariate analysis, the obtention of a CC0 score and the CA-125 concentration just before IDS were the only two statistically other different factors in the two groups (p < 0.001) (Table 1).
In our population, the discriminative ability of the Std KELIM was moderate with an area under the ROC curve of 0.67 (95% CI [0.57, 0.76]). The optimal threshold for Std KELIM defined with the Youden index was 0.61. At this cut-off point, the statistical measures of the Std KELIM performance were as follows: a sensitivity of 61%, a specificity of 72%, a positive predictive value of 77% and a negative predictive value of 51% (Fig. 2).
In our study, 113 patients met the optimal calculation of Std KELIM, i.e. using at least three CA-125 values during the first 100 days. The optimal threshold of Std KELIM in this group was found to be 0.94 with an AUC of 0.71 (95% CI [0.60, 0.81]) and a sensitivity of 43%, specificity of 94%, positive predictive value of 95% and negative predictive value of 38%.
The robustness of Std KELIM depending on the subgroups previously described was also assessed (Table 2). The subgroup defined by calculation of the Std KELIM using the first three consecutive CA-125 values had a higher discriminative capacity with an AUC of 0.73 (95% CI [0.60, 0.86])). In the CHIVA-like population, AUC was 0.71 (95% CI [0.35, 1]). The threshold defined by the Youden Index was 0.41 in this case.
|
SE
|
SP
|
PPV
|
NPV
|
YI
|
AUC
|
THRESHOLD*
|
Table 2
Discriminative ability of CA-125 and Std KELIM regarding platinum resistance
CA-125 Normalisation AFTER NAC (≤ 35 UI/mL)
|
After 3 cures
|
0.39
|
0.80
|
0.80
|
0.40
|
0.19
|
|
|
After 4 cures
|
0.57
|
0.70
|
0.80
|
0.45
|
0.27
|
|
|
StD KELIM ON OVERALL POPULATION (N = 152)
|
Optimal threshold according to the YI
|
0.61
|
0.72
|
0.77
|
0.51
|
0.29
|
0.67
[0.57–0.76]
|
0.61
|
Sensitivity analysis for the threshold
|
|
|
|
|
|
|
|
⋅ 0.5
|
0.81
|
0.37
|
0.72
|
0.50
|
0.18
|
|
|
⋅ 0.6
|
0.73
|
0.55
|
0.76
|
0.51
|
0.28
|
|
|
⋅ 0.8
|
0.51
|
0.69
|
0.76
|
0.42
|
0.20
|
|
|
⋅ 1
|
0.34
|
0.88
|
0.85
|
0.41
|
0.22
|
|
|
Sub populations
|
3 first consecutive CA-125 (n = 75)
|
0.74
|
0.67
|
0.85
|
0.50
|
0.41
|
0.73
[0.60–0.86]
|
0.60
|
CHIVA-like (n = 23)
|
0.94
|
0.60
|
0.89
|
0.75
|
0.54
|
0.71 [0.35-1]
|
0.41
|
NAC: Neoadjuvant chemotherapy; Se: sensitivity, Sp: specificity, PPV: positive predictive value, NPV: negative predictive value, YI: Youden index, AUC: area under the curve |
* Optimal threshold defined using the Youden index |
Std KELIM at the threshold in our population provides better sensitivity but less specificity for platinum-resistant recurrence (Se = 0.61, Sp = 0.72) than normalization of CA-125 (< 35 IU/mL) to third cycle of chemotherapy (Se = 0.39, Sp = 0.80), or to the fourth one (Se = 0.57, Sp = 0.70).
In the univariate analysis, Std KELIM was a significant factor of platinum-resistant recurrence, whether it was considered as a continuous covariate, where the odds ratio (OR) was 0.19 (95% CI [0.06, 0.53] p = 0.002), or a discrete covariate (p < 0.004). The lack of completeness of IDS was also a significant covariate in the univariate analyses with an OR for CC1 + versus CC0 of 4.64, (95% CI [2.27, 9.75] p < 0.001). Residual disease after IDS was however the only independent predictor in the multivariate regression model, with an OR of 3.6 (95% CI [1.6, 7.9] p = 0.001).
Platinum Resistant Recurrence Score
The ROC curve for the Platinum Resistant Recurrence Score showed good accuracy with an AUC of 0.73 (95% CI [0.64, 0.81]) (Fig. 2).
A calibration analysis of the Platinum Resistant Recurrence Score was also performed, taking into account, for each patient, the probability of platinum-resistant relapse corresponding to her IDS status, i.e. CC0 or no CC0. The calibration curve obtained showed slight but significant underestimation (p = 0.02) of the risk of platinum-resistant relapse (Fig. 3). This underestimation was present for the three terciles of our population represented. The mean absolute error in predicted probabilities was 7.1%, and the 0.9 quintile of absolute errors was 14.3%.