Table 3.
KIR association with HVL outcome, survival and NK cell function.
| Target trait | HVL outcomea | Survivalb | Number of NK cellsc | %Ki67 + NK cellsd | %CD107a-positive NK cellse | Frequency of KIR transcript sequence readsf | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Binary covariate (s) |
KIR transcript presence at pre-infection |
HVL outcome |
|||||||||||
|
Multi-KIR |
Single-KIR |
Single-KIR |
Single-KIR At ≥ 26 weeks p.i. |
Single-KIR At 8–12 weeks p.i. |
Single-KIR Longitudinal time-course |
Longitudinal level |
|||||||
| KIR | OR | (P) | OR | (P) | HR | (P) | Δ | (P) | Δ | (P) | Δ | (P) | Rank-sum test (P) |
| 1D | 0.60 | (0.656) | 1.21 | (0.735) | 1.12 | (0.777) | −0.69 | (0.993) | 0.4 | (0.947) | −1.0 | (0.755) | (0.464) |
| 3DL01 | – | – | – | – | – | – | – | – | – | – | – | – | (0.358) |
| 3DL02 | 0.01 | (0.0007) | 0.13 | (0.002) | 0.24 | (0.005) | 220.7 | (0.006) | −15.7 | (0.005) | 6.7 | (0.033) | (0.033) |
| 3DL04 | – | – | – | – | – | – | – | – | – | – | – | – | – |
| 3DL05g | 0.09 | (0.037) | 1.21 | (0.735) | 1.29 | (0.536) | −48.7 | (0.546) | 1.6 | (0.773) | 1.1 | (0.850) | (0.022) |
| 3DS05g | – | – | 2.83 | (0.068) | 1.84 | (0.127) | −118.8 | (0.128) | 7.7 | (0.174) | −4.7 | (0.172) | (0.228) |
| 3DL10g | 0.98 | (0.989) | 3.75 | (0.025) | 1.96 | (0.088) | −180.5 | (0.017) | 10.9 | (0.056) | −6.4 | (0.035) | (0.521) |
| 3DL06 | – | – | – | – | – | – | – | – | – | – | – | – | – |
| 3DL07 | 5.26 | (0.166) | 1.43 | (0.562) | 1.46 | (0.401) | −27.6 | (0.750) | 5.8 | (0.330) | 1.5 | (0.831) | (0.291) |
| 3DL08 | 17.24 | (0.211) | 1.06 | (0.921) | 1.11 | (0.798) | 15.9 | (0.841) | 0.5 | (0.933) | −1.9 | (0.456) | (0.150) |
| 3DL11 | 0.47 | (0.452) | 0.41 | (0.137) | 0.47 | (0.083) | 116.4 | (0.155) | −8.2 | (0.165) | 3.3 | (0.410) | (0.688) |
| 3DS01 | – | – | 0.15 | (0.0504) | 0.20 | (0.041) | 243.9 | (0.042) | −15.5 | (0.076) | 5.1 | (0.367) | (–)f |
| 3DS02 | 46.50 | (0.008) | 2.88 | (0.118) | 2.96 | (0.043) | −169.0 | (0.065) | 11.6 | (0.056) | −3.8 | (0.379) | (0.277) |
| 3DS03 | 0.06 | (0.238) | 1.26 | (0.711) | 1.08 | (0.861) | −53.1 | (0.548) | 3.3 | (0.610) | 1.3 | (0.695) | (0.669) |
| 3DS04 | – | – | 0.56 | (0.541) | 0.59 | (0.448) | 109.1 | (0.410) | 3.2 | (0.741) | 3.2 | (0.638) | – |
| 3DS06 | – | – | 2.50 | (0.283) | 2.07 | (0.182) | −140.9 | (0.216) | 3.5 | (0.656) | −3.4 | (0.514) | – |
| 3DSW08g | 0.10 | (0.038) | 0.21 | (0.025) | 0.29 | (0.020) | 222.8 | (0.014) | −13.2 | (0.025) | 1.6 | (0.640) | (–)f |
| 3DLW03g | – | – | 0.55 | (0.371) | 0.62 | (0.322) | 90.3 | (0.330) | −9.2 | (0.127) | −1.8 | (0.268) | (0.772) |
| 3DSW09g | 0.02 | (0.084) | 0.66 | (0.574) | 0.74 | (0.570) | 120.9 | (0.240) | −8.0 | (0.210) | −1.8 | (0.353) | (0.732) |
P values ≤0.05 in bold (significance in the multi-KIR model). Rare KIR3DL04, KIR3DL06 (n ≤ 2 with this KIR) were not analyzed. The most frequent KIR3DL01 (n = 4 without this KIR) was only analyzed for frequency of KIR transcript sequence reads.
aLogistic regression for target HVL outcome, with a single KIR or multiple KIR covariate(s): odds ratio (OR, <1 for protective KIR) and p value (P) (likelihood ratio test, χ2 statistic). ORs in the multi-KIR model are mutually adjusted within a single model [for KIRs with strata sizes ≥9 animals; excluding 3DS05, 3DLW03 due to strong correlations (see Table 2)].
bCox proportional hazards model with a single KIR covariate: hazard ratio (HR, <1 for protective KIR) and p value (P) (likelihood ratio test, χ2 statistic).
cLinear regression for target number of NK cells, with a single KIR covariate: P value (P) (t statistic) and KIR effect Δ (difference of trait means; Δ > 0 indicates a protective KIR that associates with higher NK cell numbers at ≥26 weeks p.i.).
dLinear regression for target %Ki67+ NK cells, with a single KIR covariate: P value (P) (t statistic) and KIR-effect Δ (difference of trait means; Δ < 0 indicates a protective KIR that associates with moderated NK cell proliferation at 8–12 weeks p.i.).
eLongitudinal target %CD107a-positive NK cells at pre-, acute, and chronic infection, analyzed by 1-way repeated measures ANOVA with factor KIR (n = 47 animals with data): P value (P) (ANOVA-F statistic for KIR × time interaction) and effect Δ in KIR carriers (difference between chronic and pre-infection trait means, compared to animals without KIR transcript). Note that in the cohort%CD107a-positive cells reduce over time (p < 0.0001) by −9.6%-points between pre- and chronic infection (see Figure 2). Thus, a KIR effect of 0 < Δ < 9.6 implies moderated loss-of-function (protective KIR), Δ < 0 enhanced loss-of-function (deleterious KIR).
fLongitudinal target KIR transcript frequency at pre- and acute infection (where NK cell numbers are still high), analyzed by longitudinal rank-sum test (heteroscedastic 1-way repeated measures ANOVA with factor HVL outcome on longitudinal rank-order data of the target). Analyses (a–e) above comprised the whole cohort; this analysis (f) only KIR positive animals [see Figure 4B for subsample sizes; rare KIR’s, e.g., KIR3DS01 (5 LVL, 1 HVL) and KIR3DSW08 (9 LVL, 3 HVL) were not analyzed]. P values (P) for association of KIR transcript frequency with HVL outcome; KIR3DL02 and KIR3DL05 are illustrated in Figure 4C.
gKIR3DL10–KIR3DS05–KIR3DL05 and KIR3DSW08–KIR3DLW03–KIR3DSW09 are correlated (see Table 2). The particularly strongly correlated middle KIR was excluded from the multi-KIR model.