Genito-Urinary Survey Cohort:
86
Mixed Solid Tumor Cohort:
147
External Melanoma Cohort:
157
External RCC Cohort:
28Figure 1 - Hazard Ratio Analysis of Fiber Intake
This figure is largely concerned with fiber analysis in the Genito-urinary survey cohort.
The data we are visualizing here has the following structure:
Panel A
We start with some summaries of the cohorts used here:
Panel B
Mortality Hazard for PFS by daily average fiber intake, visualized via partial smoothing spline. Ticks along the x axis represent individual patients.
cox_pfs_allMet_uv <-coxph(Surv(pfs_mo,
pod_status)~ fiber_5g_increment,
gu_cohort_data)
cat(paste(
"-----------------------------------------------\n\np value fiber in 5g increments, PFS:",
round(summary(cox_pfs_allMet_uv)$coefficients["fiber_5g_increment", "Pr(>|z|)"],digits = 3),
"\n\n-----------------------------------------------\n\n"
))-----------------------------------------------
p value fiber in 5g increments, PFS: 0.045
-----------------------------------------------summary(cox_pfs_allMet_uv)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_5g_increment,
data = gu_cohort_data)
n= 86, number of events= 55
coef exp(coef) se(coef) z Pr(>|z|)
fiber_5g_increment -0.16721 0.84602 0.08357 -2.001 0.0454 *
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
exp(coef) exp(-coef) lower .95 upper .95
fiber_5g_increment 0.846 1.182 0.7182 0.9966
Concordance= 0.597 (se = 0.039 )
Likelihood ratio test= 4.46 on 1 df, p=0.03
Wald test = 4 on 1 df, p=0.05
Score (logrank) test = 4.04 on 1 df, p=0.04Panel C
Mortality Hazard for OS by daily average fiber intake, visualized via partial smoothing spline. Ticks along the x axis represent individual patients.
cox_os_allMet_uv <-coxph(Surv(os_mo,
os_status)~ fiber_5g_increment,
gu_cohort_data)
cat(paste(
"-----------------------------------------------\n\np value fiber in 5g increments, OS:",
round(summary(cox_os_allMet_uv)$coefficients["fiber_5g_increment", "Pr(>|z|)"],digits = 3),
"\n\n-----------------------------------------------\n\n"
))-----------------------------------------------
p value fiber in 5g increments, OS: 0.048
-----------------------------------------------summary(cox_os_allMet_uv)Call:
coxph(formula = Surv(os_mo, os_status) ~ fiber_5g_increment,
data = gu_cohort_data)
n= 86, number of events= 26
coef exp(coef) se(coef) z Pr(>|z|)
fiber_5g_increment -0.2626 0.7690 0.1327 -1.979 0.0478 *
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
exp(coef) exp(-coef) lower .95 upper .95
fiber_5g_increment 0.769 1.3 0.5929 0.9974
Concordance= 0.65 (se = 0.049 )
Likelihood ratio test= 4.59 on 1 df, p=0.03
Wald test = 3.92 on 1 df, p=0.05
Score (logrank) test = 3.98 on 1 df, p=0.05Panel D
Kaplan-Meier PFS by fiber intake tertiles.
cox_pfs_allMet_mv <-coxph(Surv(pfs_mo,
pod_status)~ fiber_tertile_factor +
strata(age_cat) +
ecog_1to5_final +
cancer_RCCvsUC_NAME +
stage.simplified.factor,
gu_cohort_data
)
cat(paste(
"-----------------------------------------------\n\nWald p value top vs bottom tertile, PFS:",
round(summary(cox_pfs_allMet_mv)$coefficients["fiber_tertile_factorT3 (highest)", "Pr(>|z|)"],digits = 3),
"\n\n-----------------------------------------------\n\n"
))-----------------------------------------------
Wald p value top vs bottom tertile, PFS: 0.042
-----------------------------------------------Panel E
Kaplan-Meier OS by fiber tertile.
cox_os_allMet_mv <-coxph(Surv(os_mo,
os_status)~ fiber_tertile_factor +
strata(age_cat) +
ecog_1to5_final +
cancer_RCCvsUC_NAME +
stage.simplified.factor,
gu_cohort_data
)
cat(paste(
"-----------------------------------------------\n\nWald p value top vs bottom tertile, PFS:",
round(summary(cox_os_allMet_mv)$coefficients["fiber_tertile_factorT3 (highest)", "Pr(>|z|)"],digits = 3),
"\n\n-----------------------------------------------\n\n"
))-----------------------------------------------
Wald p value top vs bottom tertile, PFS: 0.051
-----------------------------------------------Panel F
Univariate HRs for PFS among clinically relevant subgroups in the highest versus lowest tertile of fiber consumption.
# Prep cohorts:
mUCdataFiber=filter(gu_cohort_data,cancer_type!=4)
dataUCavelumabFiber=filter(mUCdataFiber,io_regimen==5)
dataUC_NOTavelumabFiber=filter(mUCdataFiber,io_regimen!=5)
mRCdataFiber=filter(gu_cohort_data,cancer_type==4)
dataRCIpiNivoFiber=filter(mRCdataFiber,io_regimen==9 | io_regimen==17)
dataRCioTKIFiber=filter(mRCdataFiber,io_regimen==8 | io_regimen==18 | io_regimen==10 | io_regimen==11 | io_regimen==16 | io_regimen==12)
# Get pvalues:
cox_pfs_all_uv <-coxph(Surv(pfs_mo,pod_status)~ fiber_tertile_factor,
gu_cohort_data)
summary(cox_pfs_all_uv)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_tertile_factor,
data = gu_cohort_data)
n= 86, number of events= 55
coef exp(coef) se(coef) z Pr(>|z|)
fiber_tertile_factorT2 (middle) -0.6252 0.5351 0.3251 -1.923 0.0545 .
fiber_tertile_factorT3 (highest) -0.8431 0.4304 0.3370 -2.502 0.0123 *
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
exp(coef) exp(-coef) lower .95 upper .95
fiber_tertile_factorT2 (middle) 0.5351 1.869 0.2830 1.012
fiber_tertile_factorT3 (highest) 0.4304 2.324 0.2223 0.833
Concordance= 0.606 (se = 0.036 )
Likelihood ratio test= 6.85 on 2 df, p=0.03
Wald test = 7.15 on 2 df, p=0.03
Score (logrank) test = 7.49 on 2 df, p=0.02cox_pfs_RC <-coxph(Surv(pfs_mo,pod_status)~ fiber_tertile_factor,
mRCdataFiber)
summary(cox_pfs_RC)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_tertile_factor,
data = mRCdataFiber)
n= 48, number of events= 28
coef exp(coef) se(coef) z Pr(>|z|)
fiber_tertile_factorT2 (middle) -1.0867 0.3373 0.4679 -2.323 0.0202 *
fiber_tertile_factorT3 (highest) -1.0208 0.3603 0.4666 -2.188 0.0287 *
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
exp(coef) exp(-coef) lower .95 upper .95
fiber_tertile_factorT2 (middle) 0.3373 2.965 0.1348 0.8439
fiber_tertile_factorT3 (highest) 0.3603 2.775 0.1444 0.8992
Concordance= 0.659 (se = 0.05 )
Likelihood ratio test= 6.48 on 2 df, p=0.04
Wald test = 7.15 on 2 df, p=0.03
Score (logrank) test = 7.82 on 2 df, p=0.02cox_pfs_RC_IpiNivo <-coxph(Surv(pfs_mo,pod_status)~ fiber_tertile_factor,
dataRCIpiNivoFiber)
summary(cox_pfs_RC_IpiNivo)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_tertile_factor,
data = dataRCIpiNivoFiber)
n= 15, number of events= 8
coef exp(coef) se(coef) z Pr(>|z|)
fiber_tertile_factorT2 (middle) -0.7914 0.4532 0.9182 -0.862 0.389
fiber_tertile_factorT3 (highest) -0.7831 0.4570 0.8260 -0.948 0.343
exp(coef) exp(-coef) lower .95 upper .95
fiber_tertile_factorT2 (middle) 0.4532 2.207 0.07494 2.741
fiber_tertile_factorT3 (highest) 0.4570 2.188 0.09054 2.307
Concordance= 0.658 (se = 0.098 )
Likelihood ratio test= 1.04 on 2 df, p=0.6
Wald test = 1.13 on 2 df, p=0.6
Score (logrank) test = 1.19 on 2 df, p=0.6cox_pfs_RC_TKI_IO <-coxph(Surv(pfs_mo,pod_status)~ fiber_tertile_factor,
dataRCioTKIFiber)
summary(cox_pfs_RC_TKI_IO)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_tertile_factor,
data = dataRCioTKIFiber)
n= 24, number of events= 12
coef exp(coef) se(coef) z Pr(>|z|)
fiber_tertile_factorT2 (middle) -1.2070 0.2991 0.7127 -1.693 0.0904 .
fiber_tertile_factorT3 (highest) -1.1779 0.3079 0.7124 -1.653 0.0983 .
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
exp(coef) exp(-coef) lower .95 upper .95
fiber_tertile_factorT2 (middle) 0.2991 3.343 0.07398 1.209
fiber_tertile_factorT3 (highest) 0.3079 3.248 0.07621 1.244
Concordance= 0.637 (se = 0.089 )
Likelihood ratio test= 3.16 on 2 df, p=0.2
Wald test = 3.72 on 2 df, p=0.2
Score (logrank) test = 4.17 on 2 df, p=0.1cox_pfs_UC <-coxph(Surv(pfs_mo,pod_status)~ fiber_tertile_factor,
mUCdataFiber)
summary(cox_pfs_UC)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_tertile_factor,
data = mUCdataFiber)
n= 38, number of events= 27
coef exp(coef) se(coef) z Pr(>|z|)
fiber_tertile_factorT2 (middle) 0.2154 1.2404 0.4513 0.477 0.633
fiber_tertile_factorT3 (highest) -0.5447 0.5800 0.4891 -1.114 0.265
exp(coef) exp(-coef) lower .95 upper .95
fiber_tertile_factorT2 (middle) 1.24 0.8062 0.5122 3.004
fiber_tertile_factorT3 (highest) 0.58 1.7240 0.2224 1.513
Concordance= 0.606 (se = 0.06 )
Likelihood ratio test= 2.47 on 2 df, p=0.3
Wald test = 2.32 on 2 df, p=0.3
Score (logrank) test = 2.41 on 2 df, p=0.3cox_pfs_UC_NOTavelumab <-coxph(Surv(pfs_mo,pod_status)~ fiber_tertile_factor,
dataUC_NOTavelumabFiber)
summary(cox_pfs_UC_NOTavelumab)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_tertile_factor,
data = dataUC_NOTavelumabFiber)
n= 24, number of events= 19
coef exp(coef) se(coef) z Pr(>|z|)
fiber_tertile_factorT2 (middle) 0.5038 1.6550 0.5669 0.889 0.374
fiber_tertile_factorT3 (highest) -0.1227 0.8845 0.5620 -0.218 0.827
exp(coef) exp(-coef) lower .95 upper .95
fiber_tertile_factorT2 (middle) 1.6550 0.6042 0.5448 5.027
fiber_tertile_factorT3 (highest) 0.8845 1.1306 0.2940 2.661
Concordance= 0.6 (se = 0.086 )
Likelihood ratio test= 1.03 on 2 df, p=0.6
Wald test = 1.11 on 2 df, p=0.6
Score (logrank) test = 1.13 on 2 df, p=0.6cox_pfs_UC_avelumab <-coxph(Surv(pfs_mo,pod_status)~ fiber_tertile_factor,
dataUCavelumabFiber)
summary(cox_pfs_UC_avelumab)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_tertile_factor,
data = dataUCavelumabFiber)
n= 14, number of events= 8
coef exp(coef) se(coef) z Pr(>|z|)
fiber_tertile_factorT2 (middle) 0.4244 1.5286 0.8708 0.487 0.626
fiber_tertile_factorT3 (highest) -0.7228 0.4854 1.0090 -0.716 0.474
exp(coef) exp(-coef) lower .95 upper .95
fiber_tertile_factorT2 (middle) 1.5286 0.6542 0.27737 8.424
fiber_tertile_factorT3 (highest) 0.4854 2.0602 0.06718 3.507
Concordance= 0.662 (se = 0.081 )
Likelihood ratio test= 1.89 on 2 df, p=0.4
Wald test = 1.74 on 2 df, p=0.4
Score (logrank) test = 1.89 on 2 df, p=0.4forestplot_data <- tibble::tibble(mean = c(0.4304,0.3603, 0.4570, 0.3079, 0.5800, 0.8845, 0.4854),
lower = c(0.2223, 0.1444, 0.09054, 0.07621, 0.2224, 0.2940, 0.06718),
upper = c(0.833, 0.8992, 2.307, 1.244, 1.513, 2.661, 3.507),
study = c("All","RCC", "RCC ipi-nivo", "RCC TKI-ICB",
"UC", "UC non-maintenance ICB", "UC maintenance ICB"),
n = c("86","48","15","24","38","24","14"),
events = c("55","28","8","12","27","19","8"),
HR = c("0.43","0.36", "0.46", "0.31", "0.58", "0.88", "0.49"))
forestplot_data |>
forestplot(labeltext = c(study, n, events, HR),
clip = c(0.09, 2.5),
xlog = TRUE,
xlab = "Hazard ratio for PFS (95% CI)",
xticks = c(0.1, 0.25, 0.5, 1.0, 2.5),
title = "Univariable HRs for PFS comparing highest vs. lowest tertile of fiber intake",
txt_gp = fpTxtGp(xlab = grid::gpar(fontsize = 25),
ticks = grid::gpar(fontsize = 25))) |>
fp_set_style(box = "royalblue",
line = "darkblue",
summary = "royalblue") |>
fp_add_header(study = c("Subgroup"),
n = c("n"),
events = c("Events"),
HR = c("HR"))
Panel G
Multivariate HR for PFS adjusted for age, diagnosis, and performance status.
cox_pfs_all <-coxph(Surv(pfs_mo,pod_status)~ fiber_tertile_factor + strata(age_cat) +
ecog_1to5_final + cancer_RCCvsUC_NAME + stage.simplified.factor,
gu_cohort_data)
summary(cox_pfs_all)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_tertile_factor +
strata(age_cat) + ecog_1to5_final + cancer_RCCvsUC_NAME +
stage.simplified.factor, data = gu_cohort_data)
n= 86, number of events= 55
coef exp(coef) se(coef) z Pr(>|z|)
fiber_tertile_factorT2 (middle) -0.5635 0.5692 0.3459 -1.629 0.1033
fiber_tertile_factorT3 (highest) -0.7061 0.4936 0.3464 -2.038 0.0415 *
ecog_1to5_final 0.3754 1.4556 0.3405 1.102 0.2703
cancer_RCCvsUC_NAMEUC 0.5439 1.7228 0.3003 1.811 0.0701 .
stage.simplified.factorStage IV 0.5465 1.7272 1.0429 0.524 0.6003
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
exp(coef) exp(-coef) lower .95 upper .95
fiber_tertile_factorT2 (middle) 0.5692 1.7568 0.2890 1.1213
fiber_tertile_factorT3 (highest) 0.4936 2.0260 0.2503 0.9733
ecog_1to5_final 1.4556 0.6870 0.7468 2.8370
cancer_RCCvsUC_NAMEUC 1.7228 0.5805 0.9563 3.1037
stage.simplified.factorStage IV 1.7272 0.5790 0.2237 13.3374
Concordance= 0.673 (se = 0.037 )
Likelihood ratio test= 13.69 on 5 df, p=0.02
Wald test = 13.66 on 5 df, p=0.02
Score (logrank) test = 14.5 on 5 df, p=0.01cox_pfs_RC <-coxph(Surv(pfs_mo,pod_status)~ fiber_tertile_factor + ageby10 + ecog_1to5_final,
mRCdataFiber) #could not adjust for stage because all patients with RCC were stage IV
summary(cox_pfs_RC)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_tertile_factor +
ageby10 + ecog_1to5_final, data = mRCdataFiber)
n= 48, number of events= 28
coef exp(coef) se(coef) z Pr(>|z|)
fiber_tertile_factorT2 (middle) -1.4607 0.2321 0.5419 -2.695 0.00703 **
fiber_tertile_factorT3 (highest) -0.8362 0.4334 0.5012 -1.669 0.09521 .
ageby10 0.6172 1.8538 0.2150 2.871 0.00409 **
ecog_1to5_final 0.5548 1.7416 0.4812 1.153 0.24890
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
exp(coef) exp(-coef) lower .95 upper .95
fiber_tertile_factorT2 (middle) 0.2321 4.3090 0.08023 0.6713
fiber_tertile_factorT3 (highest) 0.4334 2.3076 0.16228 1.1572
ageby10 1.8538 0.5394 1.21634 2.8253
ecog_1to5_final 1.7416 0.5742 0.67822 4.4724
Concordance= 0.735 (se = 0.046 )
Likelihood ratio test= 16.38 on 4 df, p=0.003
Wald test = 16.52 on 4 df, p=0.002
Score (logrank) test = 18.19 on 4 df, p=0.001cox_pfs_UC <-coxph(Surv(pfs_mo,pod_status)~ fiber_tertile_factor + strata(age_cat) + ecog_1to5_final + stage.simplified.factor,
mUCdataFiber)
summary(cox_pfs_UC)Call:
coxph(formula = Surv(pfs_mo, pod_status) ~ fiber_tertile_factor +
strata(age_cat) + ecog_1to5_final + stage.simplified.factor,
data = mUCdataFiber)
n= 38, number of events= 27
coef exp(coef) se(coef) z Pr(>|z|)
fiber_tertile_factorT2 (middle) 0.32429 1.38305 0.51477 0.630 0.529
fiber_tertile_factorT3 (highest) -0.35545 0.70086 0.50416 -0.705 0.481
ecog_1to5_final 0.03936 1.04014 0.56864 0.069 0.945
stage.simplified.factorStage IV 0.51706 1.67709 1.08210 0.478 0.633
exp(coef) exp(-coef) lower .95 upper .95
fiber_tertile_factorT2 (middle) 1.3830 0.7230 0.5043 3.793
fiber_tertile_factorT3 (highest) 0.7009 1.4268 0.2609 1.883
ecog_1to5_final 1.0401 0.9614 0.3412 3.170
stage.simplified.factorStage IV 1.6771 0.5963 0.2011 13.984
Concordance= 0.581 (se = 0.067 )
Likelihood ratio test= 2.44 on 4 df, p=0.7
Wald test = 2.24 on 4 df, p=0.7
Score (logrank) test = 2.35 on 4 df, p=0.7forestplot_data <- tibble::tibble(mean = c(0.4936, 0.4334, 0.70086),
lower = c(0.2503, 0.16228, 0.2609),
upper = c(0.9733, 1.1572, 1.883),
study = c("All", "RCC", "UC"),
n = c("86","48","38"),
events = c("55","28","27"),
HR = c("0.49","0.43","0.70"))
forestplot_data |>
forestplot(labeltext = c(study, n, events, HR),
clip = c(0.09, 2.5),
xlog = TRUE,
xlab = "Hazard ratio for PFS (95% CI)",
xticks = c(0.1,0.25,0.5,1.0,2.5),
title = "Multivariable HRs for PFS comparing highest vs. lowest tertile of fiber intake",
txt_gp = fpTxtGp(xlab = grid::gpar(fontsize = 25),
ticks = grid::gpar(fontsize = 25))) |>
fp_set_style(box = "royalblue",
line = "darkblue",
summary = "royalblue") |>
fp_add_header(study = c("Subgroup"),
n = c("n"),
events = c("Events"),
HR = c("HR"))