Table 3. Association between FOXA1 staining level and biochemical recurrence following salvage radiation therapy.
| Single variable analysis | Multivariable analysis | |||||
|---|---|---|---|---|---|---|
| FOXA1 staining measure & group | N | Cumulative 5-year incidence of BCR, % (95% CI) | RR (95% CI) | P-value | RR (95% CI) | P-value |
| FOXA1 H-score | ||||||
| Ordinal variable | 141 | N/A | 1.05 (0.95, 1.16) | 0.33 | 1.06 (0.96, 1.18) | 0.25 |
| ≤2 | 56 | 56.8 (41.1, 68.3) | 1.00 (reference) | N/A | 1.00 (reference) | N/A |
| >2 | 85 | 52.1 (39.9, 61.8) | 0.91 (0.60, 1.39) | 0.67 | 0.98 (0.63, 1.51) | 0.93 |
| ≤3 | 77 | 58.3 (45.4, 68.2) | 1.00 (reference) | N/A | 1.00 (reference) | N/A |
| >3 | 64 | 48.7 (34.5, 59.9) | 0.85 (0.55, 1.31) | 0.45 | 1.06 (0.67, 1.69) | 0.80 |
| ≤4 | 116 | 52.2 (41.9, 60.6) | 1.00 (reference) | N/A | 1.00 (reference) | N/A |
| >4 | 25 | 62.2 (36.8, 77.4) | 1.57 (0.93, 2.63) | 0.090 | 1.61 (0.93, 2.82) | 0.092 |
| FOXA1 staining percentage | ||||||
| Ordinal variable | 141 | N/A | 1.02 (0.86, 1.20) | 0.85 | 1.09 (0.91, 1.31) | 0.33 |
| ≤25% | 37 | 50.6 (30.6, 64.8) | 1.00 (reference) | N/A | 1.00 (reference) | N/A |
| >25% | 104 | 55.3 (44.4, 64.1) | 1.09 (0.67, 1.75) | 0.74 | 1.05 (0.64, 1.73) | 0.84 |
| ≤50% | 59 | 53.7 (38.7, 65.1) | 1.00 (reference) | N/A | 1.00 (reference) | N/A |
| >50% | 82 | 54.1 (41.6, 63.9) | 1.02 (0.67, 1.56) | 0.92 | 1.19 (0.78, 1.83) | 0.42 |
| ≤75% | 85 | 55.1 (42.9, 64.6) | 1.00 (reference) | N/A | 1.00 (reference) | N/A |
| >75% | 56 | 52.2 (36.7, 64.0) | 1.02 (0.66, 1.57) | 0.94 | 1.42 (0.89, 2.27) | 0.14 |
| FOXA1 staining intensity | ||||||
| Negative or weak | 107 | 55.8 (45.0, 64.5) | 1.00 (reference) | N/A | 1.00 (reference) | N/A |
| Moderate | 34 | 48.3 (28.1, 62.8) | 1.03 (0.63, 1.68) | 0.92 | 0.84 (0.50, 1.41) | 0.51 |
RRs, 95% CIs, and p-values result from Cox proportional hazards regression models. Single variable models were adjusted for staining batch (first or second). In addition to staining batch, multivariable models were also adjusted for any variable that differed between patients with a low and high FOXA1 H-score (i.e. H-score 1–6 vs. 7–12) with a p-value of 0.10 or lower, or any variable that was associated with BCR with a p-value of 0.10 or lower. Therefore, the variables adjusted for in the multivariable models were staining batch, time from RRP to SRT, pre-SRT PSA, pathological tumor stage, and Gleason score. Three separate dichotomizations of H-score were examined in Cox regression analysis that were defined based on the 25th percentile (H-score = 2), the 50th percentile (H-score = 3), and the 75th percentile (H-score = 4). FOXA1 staining percentage was scored as one of four categories (0%–25%, 26%–50%, 51%–75%, 76%–100%) and therefore the three possible dichotomizations of this variable were examined in Cox regression analysis. Given that there were only 6 patients with negative FOXA1 staining intensity, we combined this group with the 101 patients who had weak FOXA1 staining intensity for comparison with patients who had moderate FOXA1 intensity staining in Cox regression analysis. BCR = biochemical recurrence; RR = relative risk; CI = confidence interval.