OSCR

Intensive versus standard blood pressure control and overall brain small vessel disease burden: a post-hoc analysis of the SPRINT randomized clinical trial.

Code ↔ Paper

7 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 7 matches
  1. [1] § Methods › Trial design and participants ↔ SVD_SPRINT_Git_20260705.R, lines 1–51 · score 0.88 · glomerular filtration rate, subclinical cardiovascular disease, mmHg, bodies, medications, visit
  2. [2] § Results › Attained SBP reduction and SVD change ↔ SVD_SPRINT_Git_20260705.R, lines 1–51 · score 0.87 · systolic blood pressure, basal ganglia perivascular, Dummy variables, periventricular white matter, baseline age, free water
  3. [3] § Methods › Statistical analysis › Sensitivity analyses ↔ SVD_SPRINT_Git_20260705.R, lines 1228–1275 · score 0.70 · auxiliary variables, available baseline MRI, LCS models, ICV, race, likelihood
  4. [4] § Methods › Statistical analysis › Effect of treatment assignment on SVD burden ↔ SVD_SPRINT_Git_20260705.R, lines 683–721 · score 0.68 · square root, model implied, variance, regressions, Score, covariates
  5. [5] § Methods › Statistical analysis › SVD measurement model and invariance testing strategy ↔ SVD_SPRINT_Git_20260705.R, lines 267–307 · score 0.64 · examine conceptually plausible, DIF, MIMIC, invariance, fit, latent
  6. [6] § Results › SVD measurement model and longitudinal measurement invariance ↔ SVD_SPRINT_Git_20260705.R, lines 53–93 · score 0.55 · configural invariance model, measurement invariance, metric, scalar, longitudinal, covariance
  7. [7] § Results › Effect of treatment assignment on SVD burden ↔ SVD_SPRINT_Git_20260705.R, lines 1228–1275 · score 0.51 · available baseline MRI, change score, SVD burden, LCS model, Cohen, treatment

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

R · 1,639 lines · 56 KB · no license · 7 matches

  1. # Load required libraries
  2. libraries <- c(
  3. 'tidyverse',
  4. 'lavaan',
  5. 'semTools',
  6. 'semPlot',
  7. 'semptools',
  8. 'parallel'
  9. )
  11. invisible(lapply(libraries, require, character.only = TRUE))
  13. rm(libraries)
  15. #//----- VARIABLE DICTIONARY -----
  17. # wide0: SPRINT dataset in wide format with participants who completed baseline MRI scan that passed quality control and had complete data on SVD indicators
  18. # wide01: SPRINT dataset in wide format with participants who completed both baseline and follow-up MRI scans that passed quality control and had complete data on SVD indicators
  19. # r_pvwml.0: rescaled periventricular white matter hyperintensity volume at baseline
  20. # r_pvwml.1: rescaled periventricular white matter hyperintensity volume at follow-up
  21. # r_fw.0: rescaled mean white matter free water at baseline
  22. # r_fw.1: rescaled mean white matter free water at follow-up
  23. # r_bgepvsc.0: rescaled basal ganglia perivascular space count at baseline
  24. # r_bgepvsc.1: rescaled basal ganglia perivascular space count at follow-up
  25. # r_pvwmln.0: rescaled ROI volume-normalized periventricular white matter hyperintensity volume at baseline
  26. # r_pvwmln.1: rescaled ROI volume-normalized periventricular white matter hyperintensity volume at follow-up
  27. # r_bgepvscn.0: rescaled ROI volume-normalized basal ganglia perivascular space count at baseline
  28. # r_bgepvscn.1: rescaled ROI volume-normalized basal ganglia perivascular space count at follow-up
  29. # age.c.0: mean-centered baseline age
  30. # time_years.1: time of follow-up (in years)
  31. # treat: intensive (vs standard) BP group assignment binary indicator
  32. # female: female sex binary indicator
  33. # r_icv.0: rescaled total intracranial volume at baseline
  34. # race: dummy variable vector for race/ethnicity with "White" as reference
  35. # edu: dummy variable vector for education with "College degree" as reference
  36. # smk: dummy variable vector for smoking with "Never" as reference
  37. # polyph: dummy variable vector for polypharmacy with "<5 medications" as reference
  38. # sub_cvd: subclinical cardiovascular disease binary indicator
  39. # sbp: systolic blood pressure at baseline visit
  40. # dbp: diastolic blood pressure at baseline visit
  41. # BMI: body mass index
  42. # HDL: fasting high-density lipoprotein cholesterol
  43. # result_CO2: serum bicarbonate
  44. # egfr: estimated glomerular filtration rate
  45. # log2_umalcr: log urine albumin-to-creatinine ratio
  46. # lm_delayed1: logical memory delayed score
  47. # sbp_group_1: dummy variable indicating attained SBP change from baseline of 0-10 mmHg
  48. # sbp_group_2: dummy variable indicating attained SBP change from baseline of 10-20 mmHg
  49. # sbp_group_3: dummy variable indicating attained SBP change from baseline of ≥20 mmHg
  50. # sbp_group_n: numeric attained SBP change group variable for linear trend calculation
  51. # sbp_delta: attained SBP change from baseline (continuous variable)
  53. #//--------------------------------------------------------------------------- START OF ANALYTIC CODE ---------------------------------------------------------------------------//
  55. #//-------------------------------------------- ANALYSES 3.2: SVD MEASUREMENT MODEL AND LONGITUDINAL MEASUREMENT INVARIANCE --------------------------------------------
  57. #//----- LONGITUDINAL INVARIANCE TESTING WITH RAW SVD INDICATORS -----
  59. # Testing for configural, metric (weak), and scalar (strong) measurement invariance using longitudinal CFA models for multiple indicator data
  61. # Configural invariance model (model 1)
  62. model1 <-paste0('
  63. # Measurement model for svd in time 0
  64. svd_0 =~ 1*r_pvwml.0 + fw.0*r_fw.0 + bg.0*r_bgepvsc.0
  66. # Measurement model for svd in time 1
  67. svd_1 =~ 1*r_pvwml.1 + fw.1*r_fw.1 + bg.1*r_bgepvsc.1
  69. # Mean structure specification
  70. r_pvwml.0 ~ ipv.0*1
  71. r_fw.0 ~ ifw.0*1
  72. r_bgepvsc.0 ~ ibg.0*1
  73. r_pvwml.1 ~ ipv.1*1
  74. r_fw.1 ~ ifw.1*1
  75. r_bgepvsc.1 ~ ibg.1*1
  77. # Residual covariances
  78. r_pvwml.0 ~~ r_pvwml.1
  79. r_fw.0 ~~ r_fw.1
  80. r_bgepvsc.0 ~~ r_bgepvsc.1
  82. # Latent variable means
  83. svd_0 ~ 0*1
  84. svd_1 ~ 0*1
  85. ')
  87. # Metric (weak) invariance model (model 2)
  88. model2 <-paste0('
  89. # Measurement model for svd in time 0
  90. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  92. # Measurement model for svd in time 1
  93. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  95. # Mean structure specification
  96. r_pvwml.0 ~ ipv.0*1
  97. r_fw.0 ~ ifw.0*1
  98. r_bgepvsc.0 ~ ibg.0*1
  99. r_pvwml.1 ~ ipv.1*1
  100. r_fw.1 ~ ifw.1*1
  101. r_bgepvsc.1 ~ ibg.1*1
  103. # Residual covariances
  104. r_pvwml.0 ~~ r_pvwml.1
  105. r_fw.0 ~~ r_fw.1
  106. r_bgepvsc.0 ~~ r_bgepvsc.1
  108. # Latent variable means
  109. svd_0 ~ 0*1
  110. svd_1 ~ 0*1
  111. ')
  113. # Scalar (strong) invariance model (model 3)
  114. model3 <-('
  115. # Measurement model for svd in time 0
  116. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  118. # Measurement model for svd in time 1
  119. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  121. # Mean structure specification
  122. r_pvwml.0 ~ ipv*1
  123. r_fw.0 ~ ifw*1
  124. r_bgepvsc.0 ~ ibg*1
  125. r_pvwml.1 ~ ipv*1
  126. r_fw.1 ~ ifw*1
  127. r_bgepvsc.1 ~ ibg*1
  129. # Residual covariances
  130. r_pvwml.0 ~~ r_pvwml.1
  131. r_fw.0 ~~ r_fw.1
  132. r_bgepvsc.0 ~~ r_bgepvsc.1
  134. # Latent variable means
  135. svd_0 ~ 0*1
  136. svd_1 ~ 1
  137. ')
  139. # Fit longitudinal CFA models with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  140. fit1 <- lavaan::sem(model1, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  141. fit2 <- lavaan::sem(model2, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  142. fit3 <- lavaan::sem(model3, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  144. # Compare model fit indices
  145. fitm1 <- fitmeasures(fit1, fit.measures = c("chisq.scaled", "pvalue.scaled",
  146. "cfi.robust", "rmsea.robust", "srmr"))
  148. fitm2 <- fitmeasures(fit2, fit.measures = c("chisq.scaled", "pvalue.scaled",
  149. "cfi.robust", "rmsea.robust", "srmr"))
  151. fitm3 <- fitmeasures(fit3, fit.measures = c("chisq.scaled", "pvalue.scaled",
  152. "cfi.robust", "rmsea.robust", "srmr"))
  154. fitm_all <- as.data.frame(rbind(fitm1, fitm2, fitm3))
  155. fitm_all <- fitm_all %>% mutate(across(where(is.numeric), function(x) {round(x, 3)}))
  157. print(fitm_all) # Supplemental Table 4; Models 1, 2, and 3
  159. # Display model summaries
  160. summary(fit1, fit.measures = TRUE, standardized = TRUE)
  161. summary(fit2, fit.measures = TRUE, standardized = TRUE)
  162. summary(fit3, fit.measures = TRUE, standardized = TRUE)
  165. #//----- LONGITUDINAL INVARIANCE TESTING WITH ROI VOLUME-NORMALIZED INDICATORS -----
  167. # Testing for configural, metric (weak), and scalar (strong) measurement invariance using longitudinal CFA models for multiple indicator data
  169. # Configural invariance model (model 4)
  170. model4 <-paste0('
  171. # Measurement model for svd in time 0
  172. svd_0 =~ 1*r_pvwmln.0 + fw.0*r_fw.0 + bg.0*r_bgepvscn.0
  174. # Measurement model for svd in time 1
  175. svd_1 =~ 1*r_pvwmln.1 + fw.1*r_fw.1 + bg.1*r_bgepvscn.1
  177. # Mean structure specification
  178. r_pvwmln.0 ~ ipv.0*1
  179. r_fw.0 ~ ifw.0*1
  180. r_bgepvscn.0 ~ ibg.0*1
  181. r_pvwmln.1 ~ ipv.1*1
  182. r_fw.1 ~ ifw.1*1
  183. r_bgepvscn.1 ~ ibg.1*1
  185. # Residual covariances
  186. r_pvwmln.0 ~~ r_pvwmln.1
  187. r_fw.0 ~~ r_fw.1
  188. r_bgepvscn.0 ~~ r_bgepvscn.1
  190. # Latent variable means
  191. svd_0 ~ 0*1
  192. svd_1 ~ 0*1
  193. ')
  195. # Metric (weak) invariance model (model 5)
  196. model5 <-paste0('
  197. # Measurement model for svd in time 0
  198. svd_0 =~ 1*r_pvwmln.0 + fw*r_fw.0 + bg*r_bgepvscn.0
  200. # Measurement model for svd in time 1
  201. svd_1 =~ 1*r_pvwmln.1 + fw*r_fw.1 + bg*r_bgepvscn.1
  203. # Mean structure specification
  204. r_pvwmln.0 ~ ipv.0*1
  205. r_fw.0 ~ ifw.0*1
  206. r_bgepvscn.0 ~ ibg.0*1
  207. r_pvwmln.1 ~ ipv.1*1
  208. r_fw.1 ~ ifw.1*1
  209. r_bgepvscn.1 ~ ibg.1*1
  211. # Residual covariances
  212. r_pvwmln.0 ~~ r_pvwmln.1
  213. r_fw.0 ~~ r_fw.1
  214. r_bgepvscn.0 ~~ r_bgepvscn.1
  216. # Latent variable means
  217. svd_0 ~ 0*1
  218. svd_1 ~ 0*1
  219. ')
  221. # Scalar (strong) invariance model (model 6)
  222. model6 <-('
  223. # Measurement model for svd in time 0
  224. svd_0 =~ 1*r_pvwmln.0 + fw*r_fw.0 + bg*r_bgepvscn.0
  226. # Measurement model for svd in time 1
  227. svd_1 =~ 1*r_pvwmln.1 + fw*r_fw.1 + bg*r_bgepvscn.1
  229. # Mean structure specification
  230. r_pvwmln.0 ~ ipv*1
  231. r_fw.0 ~ ifw*1
  232. r_bgepvscn.0 ~ ibg*1
  233. r_pvwmln.1 ~ ipv*1
  234. r_fw.1 ~ ifw*1
  235. r_bgepvscn.1 ~ ibg*1
  237. # Residual covariances
  238. r_pvwmln.0 ~~ r_pvwmln.1
  239. r_fw.0 ~~ r_fw.1
  240. r_bgepvscn.0 ~~ r_bgepvscn.1
  242. # Latent variable means
  243. svd_0 ~ 0*1
  244. svd_1 ~ 1
  245. ')
  247. # Fit longitudinal CFA models with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  248. fit4 <- lavaan::sem(model4, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  249. fit5 <- lavaan::sem(model5, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  250. fit6 <- lavaan::sem(model6, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  252. # Compare model fit indices
  253. fitm4 <- fitmeasures(fit4, fit.measures = c("chisq.scaled", "pvalue.scaled",
  254. "cfi.robust", "rmsea.robust", "srmr"))
  256. fitm5 <- fitmeasures(fit5, fit.measures = c("chisq.scaled", "pvalue.scaled",
  257. "cfi.robust", "rmsea.robust", "srmr"))
  259. fitm6 <- fitmeasures(fit6, fit.measures = c("chisq.scaled", "pvalue.scaled",
  260. "cfi.robust", "rmsea.robust", "srmr"))
  262. fitm_all <- as.data.frame(rbind(fitm4, fitm5, fitm6))
  263. fitm_all <- fitm_all %>% mutate(across(where(is.numeric), function(x) {round(x, 3)}))
  265. print(fitm_all) # Supplemental Table 4; Models 4, 5, and 6
  267. # Display model summaries
  268. summary(fit4, fit.measures = TRUE, standardized = TRUE)
  269. summary(fit5, fit.measures = TRUE, standardized = TRUE)
  270. summary(fit6, fit.measures = TRUE, standardized = TRUE)
  273. #//----- MULTIPLE INDICATORS MULTIPLE CAUSES (MIMIC) MODELS TO EXAMINE CONCEPTUALLY PLAUSIBLE SOURCES OF DIFFERENTIAL ITEM FUNCTIONING (DIF) OVER TIME -----
  275. # MIMIC Model for periventricular white matter hyperintensity volume (pvwml)
  276. # Metric (weak) invariance model with both direct and indirect (through SVD) effects of age on pvwml (modeldif_wml)
  277. modeldif_wml <-paste0('
  278. # Measurement model for svd in time 0
  279. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  281. # Measurement model for svd in time 1
  282. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  284. # Mean structure specification
  285. r_pvwml.0 ~ ipv.0*1
  286. r_fw.0 ~ ifw.0*1
  287. r_bgepvsc.0 ~ ibg.0*1
  288. r_pvwml.1 ~ ipv.1*1
  289. r_fw.1 ~ ifw.1*1
  290. r_bgepvsc.1 ~ ibg.1*1
  292. # Residual covariances
  293. r_pvwml.0 ~~ r_pvwml.1
  294. r_fw.0 ~~ r_fw.1
  295. r_bgepvsc.0 ~~ r_bgepvsc.1
  297. # Latent variable means
  298. svd_0 ~ 0*1
  299. svd_1 ~ 0*1
  301. # Regressions on age
  302. svd_0 ~ age.c.0
  303. svd_1 ~ age.c.0 + time_years.1
  305. r_pvwml.0 ~ age.c.0
  306. r_pvwml.1 ~ age.c.0 + time_years.1
  307. ')
  309. # Fit longitudinal MIMIC model with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  310. fitdif_wml <- lavaan::sem(modeldif_wml, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  311. summary(fitdif_wml, fit.measures = TRUE, standardized = TRUE)
  313. # Path diagram for pvwml MIMIC model (Supplementary Figure 1; panel A)
  314. pl_mod <- semPlotModel(fitdif_wml)
  316. pl_par <- pl_mod@Pars
  318. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "int", ]), ]
  320. pl_mod@Pars <- pl_par
  322. plt <- semPaths(pl_mod, whatLabels = "est", residuals = FALSE, intStyle = "multi", sizeInt = 9, edge.label.cex = 1.2)
  324. plt$layout[1, ] <- c(-1, 0.6) #p.0
  325. plt$layout[2, ] <- c(-0.7, 0.6) #f.0
  326. plt$layout[3, ] <- c(-0.4, 0.6) #b.0
  327. plt$layout[4, ] <- c(0.2, 0.6) #p.1
  328. plt$layout[6, ] <- c(0.5, 0.6) #f.1
  329. plt$layout[7, ] <- c(0.8, 0.6) #b.1
  330. plt$layout[5, ] <- c(-1.3, -0.2) #age.c.0
  331. plt$layout[8, ] <- c(-0.1, -0.2) #dufu
  332. plt$layout[9, ] <- c(-0.7, -0.8) #s_0
  333. plt$layout[10, ] <- c(0.5, -0.8) #s_1
  335. plt$graphAttributes$Nodes$labels <- c("Perive-\nticular\nWMH\nt0", "WM\nFree\nWater\nt0", "Basal\nganglia\nPVS\nt0",
  336. "Perive-\nticular\nWMH\nt1", "Base-\nline\nage", "WM\nFree\nWater\nt1",
  337. "Basal\nganglia\nPVS\nt1", "Follow\nup\ntime",
  338. "SVD\nt0", "SVD\nt1")
  340. plt <- semptools::mark_sig(semPaths_plot = plt, object = fitdif_wml)
  342. plt$graphAttributes$Nodes$label.cex <- 2
  344. plt$graphAttributes$Edges$curve[c(7)] <- 2
  345. plt$graphAttributes$Edges$curve[16] <- -1
  347. plt$graphAttributes$Edges$label.margin[c(7:9)] <- -0.035
  349. plt$graphAttributes$Edges$edge.label.position[14] <- 0.8
  351. plt$graphAttributes$Edges$edge.label.position[17] <- 0.7
  353. plt$plotOptions$label.prop <- c(0.8, 0.8, 0.8, 0.8, 0.7, 0.8, 0.8, 0.7, 0.6, 0.6)
  355. plt$graphAttributes$Edges$color[c(1:6)] <- "coral"
  357. plt$graphAttributes$Edges$color[c(7:9)] <- "lightgreen"
  359. plt$graphAttributes$Edges$color[c(10:12)] <- "magenta2"
  361. plt$graphAttributes$Edges$color[c(13:15)] <- "skyblue"
  363. plot(plt)
  366. # MIMIC Model for mean white matter free water (fw)
  367. # Metric (weak) invariance model with both direct and indirect (through SVD) effects of age on fw (modeldif_fw)
  368. modeldif_fw <-paste0('
  369. # Measurement model for svd in time 0
  370. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  372. # Measurement model for svd in time 1
  373. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  375. # Mean structure specification
  376. r_pvwml.0 ~ ipv.0*1
  377. r_fw.0 ~ ifw.0*1
  378. r_bgepvsc.0 ~ ibg.0*1
  379. r_pvwml.1 ~ ipv.1*1
  380. r_fw.1 ~ ifw.1*1
  381. r_bgepvsc.1 ~ ibg.1*1
  383. # Residual covariances
  384. r_pvwml.0 ~~ r_pvwml.1
  385. r_fw.0 ~~ r_fw.1
  386. r_bgepvsc.0 ~~ r_bgepvsc.1
  388. # Latent variable means
  389. svd_0 ~ 0*1
  390. svd_1 ~ 0*1
  392. # Regressions on age
  393. svd_0 ~ age.c.0
  394. svd_1 ~ age.c.0 + time_years.1
  396. r_fw.0 ~ age.c.0
  397. r_fw.1 ~ age.c.0 + time_years.1
  398. ')
  400. # Fit longitudinal MIMIC model with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  401. fitdif_fw <- lavaan::sem(modeldif_fw, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  402. summary(fitdif_fw, fit.measures = TRUE, standardized = TRUE)
  404. # Path diagram for fw MIMIC model (Supplementary Figure 1; panel B)
  405. pl_mod <- semPlotModel(fitdif_fw)
  407. pl_par <- pl_mod@Pars
  409. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "int", ]), ]
  411. pl_mod@Pars <- pl_par
  413. plt <- semPaths(pl_mod, whatLabels = "est", residuals = FALSE, intStyle = "multi", sizeInt = 9, edge.label.cex = 1.2)
  415. plt$layout[1, ] <- c(-1, 0.6) #p.0
  416. plt$layout[2, ] <- c(-0.7, 0.6) #f.0
  417. plt$layout[3, ] <- c(-0.4, 0.6) #b.0
  418. plt$layout[4, ] <- c(-1.3, -0.2) #age.c.0
  419. plt$layout[5, ] <- c(0.5, 0.6) #f.1
  420. plt$layout[6, ] <- c(0.2, 0.6) #p.1
  421. plt$layout[7, ] <- c(0.8, 0.6) #b.1
  422. plt$layout[8, ] <- c(-0.1, -0.2) #dufu
  423. plt$layout[9, ] <- c(-0.7, -0.8) #s_0
  424. plt$layout[10, ] <- c(0.5, -0.8) #s_1
  426. plt$graphAttributes$Nodes$labels <- c("Perive-\nticular\nWMH\nt0", "WM\nFree\nWater\nt0", "Basal\nganglia\nPVS\nt0",
  427. "Base-\nline\nage",
  428. "WM\nFree\nWater\nt1", "Perive-\nticular\nWMH\nt1", "Basal\nganglia\nPVS\nt1",
  429. "Follow\nup\ntime",
  430. "SVD\nt0", "SVD\nt1")
  432. plt <- semptools::mark_sig(semPaths_plot = plt, object = fitdif_fw)
  434. plt$graphAttributes$Nodes$label.cex <- 2
  436. plt$graphAttributes$Edges$curve[c(8)] <- 2
  437. plt$graphAttributes$Edges$curve[16] <- -1
  439. plt$graphAttributes$Edges$label.margin[c(7:9)] <- -0.035
  441. plt$graphAttributes$Edges$edge.label.position[14] <- 0.72
  443. plt$graphAttributes$Edges$edge.label.position[17] <- 0.7
  445. plt$plotOptions$label.prop <- c(0.8, 0.8, 0.8, 0.7, 0.8, 0.8, 0.8, 0.7, 0.6, 0.6)
  447. plt$graphAttributes$Edges$color[c(1:6)] <- "coral"
  449. plt$graphAttributes$Edges$color[c(7:9)] <- "lightgreen"
  451. plt$graphAttributes$Edges$color[c(10:12)] <- "magenta2"
  453. plt$graphAttributes$Edges$color[c(13:15)] <- "skyblue"
  455. plot(plt)
  458. # MIMIC Model for basal ganglia perivascular space count (bgepvsc)
  459. # Metric (weak) invariance model with both direct and indirect (through SVD) effects of age on bgepvsc (modeldif_pvs)
  460. modeldif_pvs <-paste0('
  461. # Measurement model for svd in time 0
  462. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  464. # Measurement model for svd in time 1
  465. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  467. # Mean structure specification
  468. r_pvwml.0 ~ ipv.0*1
  469. r_fw.0 ~ ifw.0*1
  470. r_bgepvsc.0 ~ ibg.0*1
  471. r_pvwml.1 ~ ipv.1*1
  472. r_fw.1 ~ ifw.1*1
  473. r_bgepvsc.1 ~ ibg.1*1
  475. # Residual covariances
  476. r_pvwml.0 ~~ r_pvwml.1
  477. r_fw.0 ~~ r_fw.1
  478. r_bgepvsc.0 ~~ r_bgepvsc.1
  480. # Latent variable means
  481. svd_0 ~ 0*1
  482. svd_1 ~ 0*1
  484. # Regressions on age
  485. svd_0 ~ age.c.0
  486. svd_1 ~ age.c.0 + time_years.1
  488. r_bgepvsc.0 ~ age.c.0
  489. r_bgepvsc.1 ~ age.c.0 + time_years.1
  490. ')
  492. # Fit longitudinal MIMIC model with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  493. fitdif_pvs <- lavaan::sem(modeldif_pvs, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  494. summary(fitdif_pvs, fit.measures = TRUE, standardized = TRUE)
  496. # Path diagram for bgepvsc MIMIC model (Supplementary Figure 1; panel C)
  497. pl_mod <- semPlotModel(fitdif_pvs)
  499. pl_par <- pl_mod@Pars
  501. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "int", ]), ]
  503. pl_mod@Pars <- pl_par
  505. plt <- semPaths(pl_mod, whatLabels = "est", residuals = FALSE, intStyle = "multi", sizeInt = 9, edge.label.cex = 1.2)
  507. plt$layout[1, ] <- c(-1, 0.6) #p.0
  508. plt$layout[2, ] <- c(-0.7, 0.6) #f.0
  509. plt$layout[3, ] <- c(-0.4, 0.6) #b.0
  510. plt$layout[4, ] <- c(-1.3, -0.2) #age.c.0
  511. plt$layout[5, ] <- c(0.2, 0.6) #p.1
  512. plt$layout[6, ] <- c(0.8, 0.6) #b.1
  513. plt$layout[7, ] <- c(0.5, 0.6) #f.1
  514. plt$layout[8, ] <- c(-0.1, -0.2) #dufu
  515. plt$layout[9, ] <- c(-0.7, -0.8) #s_0
  516. plt$layout[10, ] <- c(0.5, -0.8) #s_1
  518. plt$graphAttributes$Nodes$labels <- c("Perive-\nticular\nWMH\nt0", "WM\nFree\nWater\nt0", "Basal\nganglia\nPVS\nt0",
  519. "Base-\nline\nage",
  520. "Perive-\nticular\nWMH\nt1", "Basal\nganglia\nPVS\nt1", "WM\nFree\nWater\nt1",
  521. "Follow\nup\ntime",
  522. "SVD\nt0", "SVD\nt1")
  524. plt <- semptools::mark_sig(semPaths_plot = plt, object = fitdif_pvs)
  526. plt$graphAttributes$Nodes$label.cex <- 2
  528. plt$graphAttributes$Edges$curve[c(9)] <- 2
  529. plt$graphAttributes$Edges$curve[16] <- -1
  531. plt$graphAttributes$Edges$label.margin[c(7:9)] <- -0.035
  533. plt$graphAttributes$Edges$edge.label.position[14] <- 0.6
  535. plt$graphAttributes$Edges$edge.label.position[17] <- 0.7
  537. plt$plotOptions$label.prop <- c(0.8, 0.8, 0.8, 0.7, 0.8, 0.8, 0.8, 0.7, 0.6, 0.6)
  539. plt$graphAttributes$Edges$color[c(1:6)] <- "coral"
  541. plt$graphAttributes$Edges$color[c(7:9)] <- "lightgreen"
  543. plt$graphAttributes$Edges$color[c(10:12)] <- "magenta2"
  545. plt$graphAttributes$Edges$color[c(13:15)] <- "skyblue"
  547. plot(plt)
  550. #//----- LONGITUDINAL INVARIANCE TESTING ACCOUNTING FOR TIME-VARYING AGE EFFECTS WITH RAW SVD INDICATORS -----
  552. # Testing for configural, metric (weak), and scalar (strong) measurement invariance using longitudinal CFA models for multiple indicator data
  554. # Configural invariance model (model 7)
  555. model7 <-paste0('
  556. r_pvwml.0 ~ age.c.0
  557. r_pvwml.1 ~ age.c.0 + time_years.1
  558. r_fw.0 ~ age.c.0
  559. r_fw.1 ~ age.c.0 + time_years.1
  560. r_bgepvsc.0 ~ age.c.0
  561. r_bgepvsc.1 ~ age.c.0 + time_years.1
  563. # Measurement model for svd in time 0
  564. svd_0 =~ 1*r_pvwml.0 + fw.0*r_fw.0 + bg.0*r_bgepvsc.0
  566. # Measurement model for svd in time 1
  567. svd_1 =~ 1*r_pvwml.1 + fw.1*r_fw.1 + bg.1*r_bgepvsc.1
  569. # Mean structure specification
  570. r_pvwml.0 ~ ipv.0*1
  571. r_fw.0 ~ ifw.0*1
  572. r_bgepvsc.0 ~ ibg.0*1
  573. r_pvwml.1 ~ ipv.1*1
  574. r_fw.1 ~ ifw.1*1
  575. r_bgepvsc.1 ~ ibg.1*1
  577. # Residual covariances
  578. r_pvwml.0 ~~ r_pvwml.1
  579. r_fw.0 ~~ r_fw.1
  580. r_bgepvsc.0 ~~ r_bgepvsc.1
  582. # Latent variable means
  583. svd_0 ~ 0*1
  584. svd_1 ~ 0*1
  585. ')
  587. # Metric (weak) invariance model (model 8)
  588. model8 <-paste0('
  589. r_pvwml.0 ~ age.c.0
  590. r_pvwml.1 ~ age.c.0 + time_years.1
  591. r_fw.0 ~ age.c.0
  592. r_fw.1 ~ age.c.0 + time_years.1
  593. r_bgepvsc.0 ~ age.c.0
  594. r_bgepvsc.1 ~ age.c.0 + time_years.1
  596. # Measurement model for svd in time 0
  597. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  599. # Measurement model for svd in time 1
  600. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  602. # Mean structure specification
  603. r_pvwml.0 ~ ipv.0*1
  604. r_fw.0 ~ ifw.0*1
  605. r_bgepvsc.0 ~ ibg.0*1
  606. r_pvwml.1 ~ ipv.1*1
  607. r_fw.1 ~ ifw.1*1
  608. r_bgepvsc.1 ~ ibg.1*1
  610. # Residual covariances
  611. r_pvwml.0 ~~ r_pvwml.1
  612. r_fw.0 ~~ r_fw.1
  613. r_bgepvsc.0 ~~ r_bgepvsc.1
  615. # Latent variable means
  616. svd_0 ~ 0*1
  617. svd_1 ~ 0*1
  618. ')
  620. # Scalar (strong) invariance model (model 9)
  621. model9 <-('
  622. r_pvwml.0 ~ age.c.0
  623. r_pvwml.1 ~ age.c.0 + time_years.1
  624. r_fw.0 ~ age.c.0
  625. r_fw.1 ~ age.c.0 + time_years.1
  626. r_bgepvsc.0 ~ age.c.0
  627. r_bgepvsc.1 ~ age.c.0 + time_years.1
  629. # Measurement model for svd in time 0
  630. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  632. # Measurement model for svd in time 1
  633. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  635. # Mean structure specification
  636. r_pvwml.0 ~ ipv*1
  637. r_fw.0 ~ ifw*1
  638. r_bgepvsc.0 ~ ibg*1
  639. r_pvwml.1 ~ ipv*1
  640. r_fw.1 ~ ifw*1
  641. r_bgepvsc.1 ~ ibg*1
  643. # Residual covariances
  644. r_pvwml.0 ~~ r_pvwml.1
  645. r_fw.0 ~~ r_fw.1
  646. r_bgepvsc.0 ~~ r_bgepvsc.1
  648. # Latent variable means
  649. svd_0 ~ 0*1
  650. svd_1 ~ 1
  651. ')
  653. # Fit longitudinal CFA models with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  654. fit7 <- lavaan::sem(model7, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  655. fit8 <- lavaan::sem(model8, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  656. fit9 <- lavaan::sem(model9, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  658. # Compare model fit indices
  659. fitm7 <- fitmeasures(fit7, fit.measures = c("chisq.scaled", "pvalue.scaled",
  660. "cfi.robust", "rmsea.robust", "srmr"))
  662. fitm8 <- fitmeasures(fit8, fit.measures = c("chisq.scaled", "pvalue.scaled",
  663. "cfi.robust", "rmsea.robust", "srmr"))
  665. fitm9 <- fitmeasures(fit9, fit.measures = c("chisq.scaled", "pvalue.scaled",
  666. "cfi.robust", "rmsea.robust", "srmr"))
  668. fitm_all <- as.data.frame(rbind(fitm7, fitm8, fitm9))
  669. fitm_all <- fitm_all %>% mutate(across(where(is.numeric), function(x) {round(x, 3)}))
  671. print(fitm_all) # Supplemental Table 4; Models 7, 8, and 9
  673. # Display model summaries
  674. summary(fit7, fit.measures = TRUE, standardized = TRUE)
  675. summary(fit8, fit.measures = TRUE, standardized = TRUE)
  676. summary(fit9, fit.measures = TRUE, standardized = TRUE)
  679. #//----- PARAMETER ESTIMATES FOR THE COVARIANCE AND MEAN STRUCTURE OF THE SVD MEASUREMENT MODEL -----
  681. # Supplemental Tables 5 and 6
  683. # Model summary
  684. sum <- summary(fit9, fit.measures = TRUE, standardized = TRUE, rsquare = TRUE)
  686. # Model diagnostics
  687. lavInspect(fit9, "sampstat") # sample covariance matrix
  688. lavInspect(fit9, "implied") # model-implied covariance matrix
  689. lavInspect(fit9, "resid") # difference between observed and model-implied covariance matrix (unstandardized and unscaled model residuals)
  690. lavResiduals(fit9, type = "cor.bentler")$cov # unstandardized model residuals after transformation to correlation matrix and rescaling (by dividing the elements by the square roots of the corresponding variances of the observed covariance matrix)
  691. lavResiduals(fit9, type = "cor.bentler")$cov.z # standardized model residuals after transformation to correlation matrix and rescaling (by dividing the elements by the square roots of the corresponding variances of the observed covariance matrix)
  692. modindices(fit9, sort. = TRUE, standardized = TRUE) # sorted (from largest to smallest) model modification indices
  693. lavTestScore(fit9, cumulative = TRUE) # Score test (or Lagrange Multiplier test) for releasing one or more fixed or constrained parameters in model
  695. # Display baseline and follow-up loadings
  696. load <- sum$pe[sum$pe$op == "=~", c("rhs", "est", "se", "std.all")]
  697. load <- load %>% mutate(across(where(is.numeric), function(x) {formatC(x, format = "f", digits = 3)}))
  698. print(load)
  700. # Display baseline and follow-up age regression coefficients
  701. reg <- sum$pe[sum$pe$op == "~", c("lhs", "rhs", "est", "se", "std.all")]
  702. reg <- reg %>% mutate(across(where(is.numeric), function(x) {formatC(x, format = "f", digits = 3)}))
  703. print(reg)
  705. # Display baseline and follow-up residual (error) variances
  706. er <- sum$pe[((sum$pe$op == "~~") & (sum$pe$lhs == sum$pe$rhs)), c("lhs", "rhs", "est", "se", "std.all")]
  707. er <- er %>% mutate(across(where(is.numeric), function(x) {formatC(x, format = "f", digits = 3)}))
  708. print(er)
  710. # Display factor covariance
  711. cov <- sum$pe[((sum$pe$op == "~~") & (sum$pe$lhs != sum$pe$rhs)), c("lhs", "op", "rhs", "est", "se", "std.all")]
  712. cov <- cov %>% mutate(across(where(is.numeric), function(x) {formatC(x, format = "f", digits = 3)}))
  713. print(cov)
  715. # Display intercepts
  716. int <- sum$pe[(sum$pe$op == "~1"), c("lhs", "op", "rhs", "est", "se", "std.all")]
  717. int <- int %>% mutate(across(where(is.numeric), function(x) {formatC(x, format = "f", digits = 3)}))
  718. print(int)
  721. #//----- LONGITUDINAL INVARIANCE TESTING ACCOUNTING FOR TIME-VARYING AGE EFFECTS WITH ROI VOLUME-NORMALIZED SVD INDICATORS -----
  723. # Testing for configural, metric (weak), and scalar (strong) measurement invariance using longitudinal CFA models for multiple indicator data
  725. # Configural invariance model (model 10)
  726. model10 <-paste0('
  727. r_pvwmln.0 ~ age.c.0
  728. r_pvwmln.1 ~ age.c.0 + time_years.1
  729. r_fw.0 ~ age.c.0
  730. r_fw.1 ~ age.c.0 + time_years.1
  731. r_bgepvscn.0 ~ age.c.0
  732. r_bgepvscn.1 ~ age.c.0 + time_years.1
  734. # Measurement model for svd in time 0
  735. svd_0 =~ 1*r_pvwmln.0 + fw.0*r_fw.0 + bg.0*r_bgepvscn.0
  737. # Measurement model for svd in time 1
  738. svd_1 =~ 1*r_pvwmln.1 + fw.1*r_fw.1 + bg.1*r_bgepvscn.1
  740. # Mean structure specification
  741. r_pvwmln.0 ~ ipv.0*1
  742. r_fw.0 ~ ifw.0*1
  743. r_bgepvscn.0 ~ ibg.0*1
  744. r_pvwmln.1 ~ ipv.1*1
  745. r_fw.1 ~ ifw.1*1
  746. r_bgepvscn.1 ~ ibg.1*1
  748. # Residual covariances
  749. r_pvwmln.0 ~~ r_pvwmln.1
  750. r_fw.0 ~~ r_fw.1
  751. r_bgepvscn.0 ~~ r_bgepvscn.1
  753. # Latent variable means
  754. svd_0 ~ 0*1
  755. svd_1 ~ 0*1
  756. ')
  758. # Metric (weak) invariance model (model 11)
  759. model11 <-paste0('
  760. r_pvwmln.0 ~ age.c.0
  761. r_pvwmln.1 ~ age.c.0 + time_years.1
  762. r_fw.0 ~ age.c.0
  763. r_fw.1 ~ age.c.0 + time_years.1
  764. r_bgepvscn.0 ~ age.c.0
  765. r_bgepvscn.1 ~ age.c.0 + time_years.1
  767. # Measurement model for svd in time 0
  768. svd_0 =~ 1*r_pvwmln.0 + fw*r_fw.0 + bg*r_bgepvscn.0
  770. # Measurement model for svd in time 1
  771. svd_1 =~ 1*r_pvwmln.1 + fw*r_fw.1 + bg*r_bgepvscn.1
  773. # Mean structure specification
  774. r_pvwmln.0 ~ ipv.0*1
  775. r_fw.0 ~ ifw.0*1
  776. r_bgepvscn.0 ~ ibg.0*1
  777. r_pvwmln.1 ~ ipv.1*1
  778. r_fw.1 ~ ifw.1*1
  779. r_bgepvscn.1 ~ ibg.1*1
  781. # Residual covariances
  782. r_pvwmln.0 ~~ r_pvwmln.1
  783. r_fw.0 ~~ r_fw.1
  784. r_bgepvscn.0 ~~ r_bgepvscn.1
  786. # Latent variable means
  787. svd_0 ~ 0*1
  788. svd_1 ~ 0*1
  789. ')
  791. # Scalar (strong) invariance model (model 12)
  792. model12 <-('
  793. r_pvwmln.0 ~ age.c.0
  794. r_pvwmln.1 ~ age.c.0 + time_years.1
  795. r_fw.0 ~ age.c.0
  796. r_fw.1 ~ age.c.0 + time_years.1
  797. r_bgepvscn.0 ~ age.c.0
  798. r_bgepvscn.1 ~ age.c.0 + time_years.1
  800. # Measurement model for svd in time 0
  801. svd_0 =~ 1*r_pvwmln.0 + fw*r_fw.0 + bg*r_bgepvscn.0
  803. # Measurement model for svd in time 1
  804. svd_1 =~ 1*r_pvwmln.1 + fw*r_fw.1 + bg*r_bgepvscn.1
  806. # Mean structure specification
  807. r_pvwmln.0 ~ ipv*1
  808. r_fw.0 ~ ifw*1
  809. r_bgepvscn.0 ~ ibg*1
  810. r_pvwmln.1 ~ ipv*1
  811. r_fw.1 ~ ifw*1
  812. r_bgepvscn.1 ~ ibg*1
  814. # Residual covariances
  815. r_pvwmln.0 ~~ r_pvwmln.1
  816. r_fw.0 ~~ r_fw.1
  817. r_bgepvscn.0 ~~ r_bgepvscn.1
  819. # Latent variable means
  820. svd_0 ~ 0*1
  821. svd_1 ~ 1
  822. ')
  824. # Fit longitudinal CFA models with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  825. fit10 <- lavaan::sem(model10, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  826. fit11 <- lavaan::sem(model11, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  827. fit12 <- lavaan::sem(model12, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  829. # Compare model fit indices
  830. fitm10 <- fitmeasures(fit10, fit.measures = c("chisq.scaled", "pvalue.scaled",
  831. "cfi.robust", "rmsea.robust", "srmr"))
  833. fitm11 <- fitmeasures(fit11, fit.measures = c("chisq.scaled", "pvalue.scaled",
  834. "cfi.robust", "rmsea.robust", "srmr"))
  836. fitm12 <- fitmeasures(fit12, fit.measures = c("chisq.scaled", "pvalue.scaled",
  837. "cfi.robust", "rmsea.robust", "srmr"))
  839. fitm_all <- as.data.frame(rbind(fitm10, fitm11, fitm12))
  840. fitm_all <- fitm_all %>% mutate(across(where(is.numeric), function(x) {round(x, 3)}))
  842. print(fitm_all) # Supplemental Table 4; Models 10, 11, and 12
  844. # Display model summaries
  845. summary(fit10, fit.measures = TRUE, standardized = TRUE)
  846. summary(fit11, fit.measures = TRUE, standardized = TRUE)
  847. summary(fit12, fit.measures = TRUE, standardized = TRUE)
  850. #//----- MULTIPLE-INDICATOR LATENT CHANGE SCORE (LCS) MODELS -----
  852. # LCS model equivalent to longitudinal CFA model 9
  853. model13 <-paste0('
  854. r_pvwml.0 ~ age.c.0
  855. r_pvwml.1 ~ age.c.0 + time_years.1
  856. r_fw.0 ~ age.c.0
  857. r_fw.1 ~ age.c.0 + time_years.1
  858. r_bgepvsc.0 ~ age.c.0
  859. r_bgepvsc.1 ~ age.c.0 + time_years.1
  861. # Measurement model for svd in time 0
  862. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  864. # Measurement model for svd in time 1
  865. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  867. # Mean structure specification
  868. r_pvwml.0 ~ 0*1
  869. r_fw.0 ~ ifw*1
  870. r_bgepvsc.0 ~ ibg*1
  871. r_pvwml.1 ~ 0*1
  872. r_fw.1 ~ ifw*1
  873. r_bgepvsc.1 ~ ibg*1
  875. # Residual covariances
  876. r_pvwml.0 ~~ r_pvwml.1
  877. r_fw.0 ~~ r_fw.1
  878. r_bgepvsc.0 ~~ r_bgepvsc.1
  880. # Latent change
  881. svd_1 ~ 1*svd_0
  882. dC =~ 1*svd_1
  884. # Latent variable mean structure
  885. svd_0 ~ 1
  886. svd_1 ~ 0*1
  887. dC ~ 1
  889. # Latent variable covariance structure
  890. svd_0 ~~ svd_0
  891. svd_1 ~~ 0*svd_1
  892. dC ~~ dC
  893. dC ~~ svd_0
  894. ')
  896. # Fit LCS model with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  897. fit13 <- lavaan::sem(model13, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  898. summary(fit13, fit.measures = TRUE, standardized = TRUE)
  900. # Compare models (sanity check)
  901. lavTestLRT(fit9, fit13)
  903. # Compare model fit indices (sanity check)
  904. fitm9 <- fitmeasures(fit9, fit.measures = c("chisq.scaled", "pvalue.scaled",
  905. "cfi.robust", "tli.robust",
  906. "rmsea.robust", "srmr"))
  908. fitm13 <- fitmeasures(fit13, fit.measures = c("chisq.scaled", "pvalue.scaled",
  909. "cfi.robust", "tli.robust",
  910. "rmsea.robust", "srmr"))
  912. fitm_all <- as.data.frame(rbind(fitm9, fitm13))
  913. fitm_all <- fitm_all %>% mutate(across(where(is.numeric), function(x) {round(x, 3)}))
  915. print(fitm_all)
  918. #//-------------------------------------------- ANALYSES 3.3: EFFECT OF TREATMENT ASSIGNMENT ON SVD BURDEN --------------------------------------------
  920. #//----- COMPUTE UNADJUSTED TREATMENT EFFECT -----
  921. # LCS model with treatment arm as latent change score predictor (unadjusted)
  922. model1 <-paste0('
  923. r_pvwml.0 ~ age.c.0
  924. r_pvwml.1 ~ age.c.0 + time_years.1
  925. r_fw.0 ~ age.c.0
  926. r_fw.1 ~ age.c.0 + time_years.1
  927. r_bgepvsc.0 ~ age.c.0
  928. r_bgepvsc.1 ~ age.c.0 + time_years.1
  930. # Measurement model for svd in time 0
  931. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  933. # Measurement model for svd in time 1
  934. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  936. # Mean structure specification
  937. r_pvwml.0 ~ 0*1
  938. r_fw.0 ~ ifw*1
  939. r_bgepvsc.0 ~ ibg*1
  940. r_pvwml.1 ~ 0*1
  941. r_fw.1 ~ ifw*1
  942. r_bgepvsc.1 ~ ibg*1
  944. # Residual covariances
  945. r_pvwml.0 ~~ r_pvwml.1
  946. r_fw.0 ~~ r_fw.1
  947. r_bgepvsc.0 ~~ r_bgepvsc.1
  949. # Latent change
  950. svd_1 ~ 1*svd_0
  951. dC =~ 1*svd_1
  953. # Latent variable mean structure
  954. svd_0 ~ a0*1
  955. svd_1 ~ 0*1
  956. dC ~ 1 + a*treat
  958. # Latent variable covariance structure
  959. svd_0 ~~ svd_0
  960. svd_1 ~~ 0*svd_1
  961. dC ~~ b*dC
  962. dC ~~ svd_0
  964. # Standardized mean difference (Cohens d) and % change relative to mean baseline SVD burden
  965. cohen_d := a/sqrt(b)
  966. prop := a/a0
  967. ')
  969. # Fit LCS model with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  970. fit1 <- lavaan::sem(model1, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  971. summary(fit1, fit.measures = TRUE, standardized = TRUE)
  973. # Inspect fit indices
  974. fitmeasures(fit1, fit.measures = c("chisq.scaled", "pvalue.scaled",
  975. "cfi.robust", "rmsea.robust", "srmr"))
  977. # Display parameter estimates of interest
  978. parameterEstimates(fit1)[parameterEstimates(fit1)$label %in%
  979. c("a", "cohen_d", "prop"),
  980. c("lhs", "est", "ci.lower", "ci.upper")]
  982. # Display latent change score variance
  983. parameterEstimates(fit1)[parameterEstimates(fit1)$label %in% "b", "est"]
  985. # Path diagram for the unadjusted treatment effect (Figure 2; panel A)
  986. pl_mod <- semPlotModel(fit1)
  988. pl_par <- pl_mod@Pars
  990. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "int" & !pl_par$rhs %in% c("dC", "svd_0"), ]), ]
  992. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "<->" & pl_par$rhs %in% c("treat"), ]), ]
  994. pl_mod@Pars <- pl_par
  996. plt <- semPaths(pl_mod, whatLabels = "est", residuals = FALSE, intStyle = "multi", sizeInt = 9, edge.label.cex = 1.2)
  998. plt$layout[1, ] <- c(-1, 0.6) #p.0
  999. plt$layout[2, ] <- c(-0.7, 0.6) #f.0
  1000. plt$layout[3, ] <- c(-0.4, 0.6) #b.0
  1001. plt$layout[4, ] <- c(0.2, 0.6) #p.1
  1002. plt$layout[5, ] <- c(0.5, 0.6) #f.1
  1003. plt$layout[6, ] <- c(0.8, 0.6) #b.1
  1004. plt$layout[7, ] <- c(0.3, -1) #trt
  1005. plt$layout[8, ] <- c(-0.7, 1.35) #age.c.0
  1006. plt$layout[9, ] <- c(0.5, 1.35) #dufu
  1007. plt$layout[10, ] <- c(-0.7, -0.5) #s_0
  1008. plt$layout[11, ] <- c(0.5, -0.5) #s_1
  1009. plt$layout[12, ] <- c(-0.2, -1) #dC
  1010. plt$layout[13, ] <- c(-1.2, -0.5) #int.s_0
  1011. plt$layout[14, ] <- c(-0.8, -1) #int.dC
  1013. plt$graphAttributes$Nodes$labels <- c("Perive-\nticular\nWMH\nt0", "WM\nFree\nWater\nt0", "Basal\nganglia\nPVS\nt0",
  1014. "Perive-\nticular\nWMH\nt1", "WM\nFree\nWater\nt1", "Basal\nganglia\nPVS\nt1",
  1015. "Intensive\nBP\nControl",
  1016. "Base-\nline\nage", "Follow\nup\ntime",
  1017. "SVD\nt0", "SVD\nt1", "SVD\nChange",
  1018. "Mean\nBaseline\nSVD", "Mean\nSVD\nChange")
  1020. plt <- semptools::mark_sig(semPaths_plot = plt, object = fit1)
  1022. plt$graphAttributes$Nodes$label.cex <- 1.6
  1024. plt$graphAttributes$Edges$curve[c(24,25)] <- 0.5
  1026. plt$graphAttributes$Edges$label.margin[c(16:18)] <- -0.035
  1027. plt$graphAttributes$Edges$label.margin[c(21,23)] <- -0.04
  1028. plt$graphAttributes$Edges$label.margin[c(22)] <- -0.03
  1030. plt$plotOptions$label.prop <- c(1, 1, 1, 1, 1, 1, 0.95, 0.85, 0.85, 0.7, 0.7, 0.85, 0.45, 0.425)
  1032. plt$graphAttributes$Edges$color[c(1,2,4,5,7,8)] <- "skyblue"
  1034. plt$graphAttributes$Edges$color[c(3,6,9)] <- "navy"
  1036. plt$graphAttributes$Edges$color[c(10:15)] <- "coral"
  1038. plt$graphAttributes$Edges$color[c(16:18)] <- "lightgreen"
  1040. plt$graphAttributes$Edges$color[c(21:22)] <- "blue"
  1042. plt$graphAttributes$Edges$color[c(23)] <- "red2"
  1044. plt$graphAttributes$Nodes$width <- c(5, 5, 5, 5, 5, 5, 7, 5, 5, 8, 8, 8, 12, 12)
  1046. plot(plt)
  1049. #//----- COMPUTE ADJUSTED TREATMENT EFFECT -----
  1050. # LCS model with treatment arm as latent change score predictor adjusted for baseline age, sex, race, and icv
  1051. model2 <-paste0('
  1052. r_pvwml.0 ~ age.c.0
  1053. r_pvwml.1 ~ age.c.0 + time_years.1
  1054. r_fw.0 ~ age.c.0
  1055. r_fw.1 ~ age.c.0 + time_years.1
  1056. r_bgepvsc.0 ~ age.c.0
  1057. r_bgepvsc.1 ~ age.c.0 + time_years.1
  1059. # Measurement model for svd in time 0
  1060. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  1062. # Measurement model for svd in time 1
  1063. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  1065. # Mean structure specification
  1066. r_pvwml.0 ~ 0*1
  1067. r_fw.0 ~ ifw*1
  1068. r_bgepvsc.0 ~ ibg*1
  1069. r_pvwml.1 ~ 0*1
  1070. r_fw.1 ~ ifw*1
  1071. r_bgepvsc.1 ~ ibg*1
  1073. # Residual covariances
  1074. r_pvwml.0 ~~ r_pvwml.1
  1075. r_fw.0 ~~ r_fw.1
  1076. r_bgepvsc.0 ~~ r_bgepvsc.1
  1078. # Latent change
  1079. svd_1 ~ 1*svd_0
  1080. dC =~ 1*svd_1
  1082. # Latent variable mean structure
  1083. svd_0 ~ a0*1 + age.c.0 + female + r_icv.0 + ', paste(c(race), collapse = " + "), '
  1084. svd_1 ~ 0*1
  1085. dC ~ 1 + a*treat + age.c.0 + female + r_icv.0 + ', paste(c(race), collapse = " + "), '
  1087. # Latent variable covariance structure
  1088. svd_0 ~~ svd_0
  1089. svd_1 ~~ 0*svd_1
  1090. dC ~~ b*dC
  1091. dC ~~ svd_0
  1093. # Standardized mean difference (Cohens d)
  1094. cohen_d := a/sqrt(b)
  1095. ')
  1097. # Fit LCS model with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  1098. fit2 <- lavaan::sem(model2, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  1099. summary(fit2, fit.measures = TRUE, standardized = TRUE)
  1101. # Inspect fit indices
  1102. fitmeasures(fit2, fit.measures = c("chisq.scaled", "pvalue.scaled",
  1103. "cfi.robust", "rmsea.robust", "srmr"))
  1105. # Display parameter estimates of interest
  1106. parameterEstimates(fit2)[parameterEstimates(fit2)$label %in%
  1107. c("a", "cohen_d"),
  1108. c("lhs", "est", "ci.lower", "ci.upper")]
  1110. # Display latent change score variance
  1111. parameterEstimates(fit2)[parameterEstimates(fit2)$label %in% "b", "est"]
  1113. # Path diagram for the adjusted treatment effect (Figure 2; panel B)
  1114. pl_mod <- semptools::drop_nodes(
  1115. object = semPlotModel(fit2),
  1116. nodes = c(race, "female", "r_icv.0"))
  1118. pl_par <- pl_mod@Pars
  1120. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "int" & !pl_par$rhs %in% c("dC", "svd_0"), ]), ]
  1122. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "<->" & pl_par$rhs %in% c("treat"), ]), ]
  1124. pl_mod@Pars <- pl_par
  1126. plt <- semPaths(pl_mod, whatLabels = "est", residuals = FALSE, intStyle = "multi", sizeInt = 9, edge.label.cex = 1.2)
  1128. plt$layout[1, ] <- c(-1, 0.6) #p.0
  1129. plt$layout[2, ] <- c(-0.7, 0.6) #f.0
  1130. plt$layout[3, ] <- c(-0.4, 0.6) #b.0
  1131. plt$layout[4, ] <- c(0.2, 0.6) #p.1
  1132. plt$layout[5, ] <- c(0.5, 0.6) #f.1
  1133. plt$layout[6, ] <- c(0.8, 0.6) #b.1
  1134. plt$layout[7, ] <- c(-0.7, 1.35) #age.c.0
  1135. plt$layout[8, ] <- c(0.3, -1) #trt
  1136. plt$layout[9, ] <- c(0.5, 1.35) #dufu
  1137. plt$layout[10, ] <- c(-0.7, -0.5) #s_0
  1138. plt$layout[11, ] <- c(0.5, -0.5) #s_1
  1139. plt$layout[12, ] <- c(-0.2, -1) #dC
  1140. plt$layout[13, ] <- c(-1.2, -0.5) #int.s_0
  1141. plt$layout[14, ] <- c(-0.8, -1) #int.dC
  1143. plt$graphAttributes$Nodes$labels <- c("Perive-\nticular\nWMH\nt0", "WM\nFree\nWater\nt0", "Basal\nganglia\nPVS\nt0",
  1144. "Perive-\nticular\nWMH\nt1", "WM\nFree\nWater\nt1", "Basal\nganglia\nPVS\nt1",
  1145. "Base-\nline\nage", "Intensive\nBP\nControl", "Follow\nup\ntime",
  1146. "SVD\nt0", "SVD\nt1", "SVD\nChange",
  1147. "Mean\nBaseline\nSVD", "Mean\nSVD\nChange")
  1149. plt <- semptools::mark_sig(semPaths_plot = plt, object = fit2)
  1151. plt$graphAttributes$Nodes$label.cex <- 1.6
  1153. plt$graphAttributes$Edges$curve[c(22)] <- -6
  1155. plt$graphAttributes$Edges$curve[c(26,27)] <- 0.5
  1157. plt$graphAttributes$Edges$label.margin[c(16:18)] <- -0.035
  1158. plt$graphAttributes$Edges$label.margin[c(21,23,24)] <- -0.04
  1159. plt$graphAttributes$Edges$label.margin[c(23)] <- -0.03
  1161. plt$graphAttributes$Edges$edge.label.position[25] <- 0.7
  1163. plt$plotOptions$label.prop <- c(1, 1, 1, 1, 1, 1, 0.85, 0.95, 0.85, 0.7, 0.7, 0.85, 0.45, 0.425)
  1165. plt$graphAttributes$Edges$color[c(1,2,4,5,7,8)] <- "skyblue"
  1167. plt$graphAttributes$Edges$color[c(3,6,9)] <- "navy"
  1169. plt$graphAttributes$Edges$color[c(10:15)] <- "coral"
  1171. plt$graphAttributes$Edges$color[c(16:18)] <- "lightgreen"
  1173. plt$graphAttributes$Edges$color[c(21,23)] <- "blue"
  1175. plt$graphAttributes$Edges$color[c(24)] <- "red2"
  1177. plt$graphAttributes$Edges$color[c(22, 25)] <- "magenta2"
  1179. plt$graphAttributes$Nodes$width <- c(5, 5, 5, 5, 5, 5, 5, 7, 5, 8, 8, 8, 12, 12)
  1181. plot(plt)
  1184. #//----- COMPUTE TREATMENT EFFECT WITH ROI VOLUME-NORMALIZED SVD INDICATORS -----
  1185. # LCS model with treatment arm as latent change score predictor (unadjusted) using ROI volume-adjusted MRI indicators
  1186. model3 <-paste0('
  1187. r_pvwmln.0 ~ age.c.0
  1188. r_pvwmln.1 ~ age.c.0 + time_years.1
  1189. r_fw.0 ~ age.c.0
  1190. r_fw.1 ~ age.c.0 + time_years.1
  1191. r_bgepvscn.0 ~ age.c.0
  1192. r_bgepvscn.1 ~ age.c.0 + time_years.1
  1194. # Measurement model for svd in time 0
  1195. svd_0 =~ 1*r_pvwmln.0 + fw*r_fw.0 + bg*r_bgepvscn.0
  1197. # Measurement model for svd in time 1
  1198. svd_1 =~ 1*r_pvwmln.1 + fw*r_fw.1 + bg*r_bgepvscn.1
  1200. # Mean structure specification
  1201. r_pvwmln.0 ~ 0*1
  1202. r_fw.0 ~ ifw*1
  1203. r_bgepvscn.0 ~ ibg*1
  1204. r_pvwmln.1 ~ 0*1
  1205. r_fw.1 ~ ifw*1
  1206. r_bgepvscn.1 ~ ibg*1
  1208. # Residual covariances
  1209. r_pvwmln.0 ~~ r_pvwmln.1
  1210. r_fw.0 ~~ r_fw.1
  1211. r_bgepvscn.0 ~~ r_bgepvscn.1
  1213. # Latent change
  1214. svd_1 ~ 1*svd_0
  1215. dC =~ 1*svd_1
  1217. # Latent variable mean structure
  1218. svd_0 ~ a0*1
  1219. svd_1 ~ 0*1
  1220. dC ~ 1 + a*treat
  1222. # Latent variable covariance structure
  1223. svd_0 ~~ svd_0
  1224. svd_1 ~~ 0*svd_1
  1225. dC ~~ b*dC
  1226. dC ~~ svd_0
  1228. # Standardized mean difference (Cohens d) and % change relative to mean baseline SVD burden
  1229. cohen_d := a/sqrt(b)
  1230. prop := a/a0
  1231. ')
  1233. # Fit LCS model with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  1234. fit3 <- lavaan::sem(model3, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  1235. summary(fit3, fit.measures = TRUE, standardized = TRUE)
  1237. # Inspect fit indices
  1238. fitmeasures(fit3, fit.measures = c("chisq.scaled", "pvalue.scaled",
  1239. "cfi.robust", "rmsea.robust", "srmr"))
  1241. # Display parameter estimates of interest
  1242. parameterEstimates(fit3)[parameterEstimates(fit3)$label %in%
  1243. c("a", "cohen_d", "prop"),
  1244. c("lhs", "est", "ci.lower", "ci.upper")]
  1246. # Display latent change score variance
  1247. parameterEstimates(fit3)[parameterEstimates(fit3)$label %in% "b", "est"]
  1250. #//----- COMPUTE TREATMENT EFFECT IN ALL PARTICIPANTS WITH AVAILABLE BASELINE MRI WITH FIML -----
  1251. # LCS model for the total sample of participants with baseline MRI with treatment arm as latent change score predictor (unadjusted) with fiml for missing follow-up MRI markers + auxiliary variables
  1253. # Create dataset for fiml
  1254. fiml <- wide0
  1256. # Base auxiliary variable set
  1257. base_set <- c('female', race, edu, polyph, smk, 'sub_cvd', 'sbp', 'dbp')
  1258. # Inspect for missing values
  1259. sapply(base_set, function(x){sum(is.na(fiml[ ,x]))})
  1261. # Extended auxiliary variable set
  1262. ext_set <- c("BMI", "HDL", "result_CO2", "egfr", "log2_umalcr", "lm_delayed1", "r_icv.0")
  1263. # Inspect for missing values
  1264. sapply(ext_set, function(x){sum(is.na(fiml[ ,x]))})
  1266. # Fit LCS model with Full Information Maximum Likelihood estimation method with auxiliary variables using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  1267. fit.aux <- auxiliary(model1, aux = c(base_set, ext_set), fun = "sem", data = fiml, estimator = "ML", missing = "FIML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  1268. summary(fit.aux, fit.measures = TRUE, standardized = TRUE)
  1270. fitmeasures(fit.aux, fit.measures = c("chisq.scaled", "pvalue.scaled",
  1271. "cfi.robust", "rmsea.robust", "srmr"))
  1273. parameterEstimates(fit.aux)[parameterEstimates(fit.aux)$label %in%
  1274. c("a", "cohen_d", "prop"),
  1275. c("lhs", "est", "ci.lower", "ci.upper")]
  1278. #//-------------------------------------------- ANALYSES 3.4: ATTAINED SBP REDUCTION AND SVD CHANGE --------------------------------------------
  1279. # LCS model with SBP delta group as latent change score predictor
  1280. model4 <-paste0('
  1281. r_pvwml.0 ~ age.c.0
  1282. r_pvwml.1 ~ age.c.0 + time_years.1
  1283. r_fw.0 ~ age.c.0
  1284. r_fw.1 ~ age.c.0 + time_years.1
  1285. r_bgepvsc.0 ~ age.c.0
  1286. r_bgepvsc.1 ~ age.c.0 + time_years.1
  1288. # Measurement model for svd in time 0
  1289. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  1291. # Measurement model for svd in time 1
  1292. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  1294. # Mean structure specification
  1295. r_pvwml.0 ~ 0*1
  1296. r_fw.0 ~ ifw*1
  1297. r_bgepvsc.0 ~ ibg*1
  1298. r_pvwml.1 ~ 0*1
  1299. r_fw.1 ~ ifw*1
  1300. r_bgepvsc.1 ~ ibg*1
  1302. # Residual covariances
  1303. r_pvwml.0 ~~ r_pvwml.1
  1304. r_fw.0 ~~ r_fw.1
  1305. r_bgepvsc.0 ~~ r_bgepvsc.1
  1307. # Latent change
  1308. svd_1 ~ 1*svd_0
  1309. dC =~ 1*svd_1
  1311. # Latent variable mean structure
  1312. svd_0 ~ a0*1
  1313. svd_1 ~ 0*1
  1314. dC ~ 1 + a1*sbp_group_1 + a2*sbp_group_2 + a3*sbp_group_3
  1316. # Latent variable covariance structure
  1317. svd_0 ~~ svd_0
  1318. svd_1 ~~ 0*svd_1
  1319. dC ~~ dC
  1320. dC ~~ svd_0
  1322. # % change relative to mean baseline SVD burden
  1323. prop_1 := a1/a0
  1324. prop_2 := a2/a0
  1325. prop_3 := a3/a0
  1326. ')
  1328. # Fit LCS model with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  1329. fit4 <- lavaan::sem(model4, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  1330. summary(fit4, fit.measures = TRUE, standardized = TRUE)
  1332. # Inspect fit indices
  1333. fitmeasures(fit4, fit.measures = c("chisq.scaled", "pvalue.scaled",
  1334. "cfi.robust", "rmsea.robust", "srmr"))
  1336. # Display parameter estimates of interest
  1337. parameterEstimates(fit4)[parameterEstimates(fit4)$label %in%
  1338. c("a1", "a2", "a3",
  1339. "prop_1", "prop_2", "prop_3"),
  1340. c("lhs", "est", "ci.lower", "ci.upper")]
  1342. # Path diagram for different SBP delta groups and SVD burden change (Figure 3)
  1343. pl_mod <- semPlotModel(fit4)
  1345. pl_par <- pl_mod@Pars
  1347. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "int" & !pl_par$rhs %in% c("dC", "svd_0"), ]), ]
  1349. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "<->" & pl_par$rhs %in%
  1350. c("sbp_group_1", "sbp_group_2", "sbp_group_3"), ]), ]
  1352. pl_mod@Pars <- pl_par
  1354. plt <- semPaths(pl_mod, whatLabels = "est", residuals = FALSE, intStyle = "multi", sizeInt = 9, edge.label.cex = 1.2)
  1356. plt$layout[1, ] <- c(-1, 0.6) #p.0
  1357. plt$layout[2, ] <- c(-0.7, 0.6) #f.0
  1358. plt$layout[3, ] <- c(-0.4, 0.6) #b.0
  1359. plt$layout[4, ] <- c(0.2, 0.6) #p.1
  1360. plt$layout[5, ] <- c(0.5, 0.6) #f.1
  1361. plt$layout[6, ] <- c(0.8, 0.6) #b.1
  1363. plt$layout[7, ] <- c(0.8, -0.7) #sbp_1
  1364. plt$layout[8, ] <- c(0.8, -1.0) #sbp_2
  1365. plt$layout[9, ] <- c(0.8, -1.3) #sbp_3
  1367. plt$layout[10, ] <- c(-0.7, 1.35) #age.c.0
  1368. plt$layout[11, ] <- c(0.5, 1.35) #dufu
  1369. plt$layout[12, ] <- c(-0.7, -0.5) #s_0
  1370. plt$layout[13, ] <- c(0.5, -0.5) #s_1
  1371. plt$layout[14, ] <- c(-0.2, -1) #dC
  1372. plt$layout[15, ] <- c(-1.2, -0.5) #int.s_0
  1373. plt$layout[16, ] <- c(-0.8, -1) #int.dC
  1375. plt$graphAttributes$Nodes$labels <- c("Perive-\nticular\nWMH\nt0", "WM\nFree\nWater\nt0", "Basal\nganglia\nPVS\nt0",
  1376. "Perive-\nticular\nWMH\nt1", "WM\nFree\nWater\nt1", "Basal\nganglia\nPVS\nt1",
  1377. "SBP\nDrop\n0 to 10", "SBP\nDrop\n10 to 20", "SBP\nDrop\n≥ 20",
  1378. "Base-\nline\nage", "Follow\nup\ntime",
  1379. "SVD\nt0", "SVD\nt1", "SVD\nChange",
  1380. "Mean\nBaseline\nSVD", "Mean\nSVD\nChange")
  1382. plt <- semptools::mark_sig(semPaths_plot = plt, object = fit4)
  1384. plt$graphAttributes$Nodes$label.cex <- 1.6
  1386. plt$graphAttributes$Edges$curve[c(26,27)] <- 0.5
  1388. plt$graphAttributes$Edges$label.margin[c(16:18)] <- -0.035
  1389. plt$graphAttributes$Edges$label.margin[c(22, 23)] <- -0.02
  1390. plt$graphAttributes$Edges$label.margin[c(21, 24, 25)] <- -0.03
  1392. plt$plotOptions$label.prop <- c(1, 1, 1,
  1393. 1, 1, 1,
  1394. 0.9, 1, 0.9,
  1395. 0.85, 0.85,
  1396. 0.7, 0.7, 0.85,
  1397. 0.45, 0.425)
  1399. plt$graphAttributes$Edges$color[c(1,2,4,5,7,8)] <- "skyblue"
  1401. plt$graphAttributes$Edges$color[c(3,6,9)] <- "navy"
  1403. plt$graphAttributes$Edges$color[c(10:15)] <- "coral"
  1405. plt$graphAttributes$Edges$color[c(16:18)] <- "lightgreen"
  1407. plt$graphAttributes$Edges$color[c(21:22)] <- "blue"
  1409. plt$graphAttributes$Edges$color[c(23:25)] <- "red2"
  1411. plt$graphAttributes$Nodes$width <- c(5, 5, 5,
  1412. 5, 5, 5,
  1413. 5, 5, 5,
  1414. 5, 5,
  1415. 8, 8, 8,
  1416. 12, 12)
  1418. plot(plt)
  1421. #//----- TREND TEST -----
  1422. # LCS model with BP delta group as latent change score predictor - trend test
  1423. model5 <-paste0('
  1424. r_pvwml.0 ~ age.c.0
  1425. r_pvwml.1 ~ age.c.0 + time_years.1
  1426. r_fw.0 ~ age.c.0
  1427. r_fw.1 ~ age.c.0 + time_years.1
  1428. r_bgepvsc.0 ~ age.c.0
  1429. r_bgepvsc.1 ~ age.c.0 + time_years.1
  1431. # Measurement model for svd in time 0
  1432. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  1434. # Measurement model for svd in time 1
  1435. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  1437. # Mean structure specification
  1438. r_pvwml.0 ~ 0*1
  1439. r_fw.0 ~ ifw*1
  1440. r_bgepvsc.0 ~ ibg*1
  1441. r_pvwml.1 ~ 0*1
  1442. r_fw.1 ~ ifw*1
  1443. r_bgepvsc.1 ~ ibg*1
  1445. # Residual covariances
  1446. r_pvwml.0 ~~ r_pvwml.1
  1447. r_fw.0 ~~ r_fw.1
  1448. r_bgepvsc.0 ~~ r_bgepvsc.1
  1450. # Latent change
  1451. svd_1 ~ 1*svd_0
  1452. dC =~ 1*svd_1
  1454. # Latent variable mean structure
  1455. svd_0 ~ 1
  1456. svd_1 ~ 0*1
  1457. dC ~ 1 + sbp_group_n
  1459. # Latent variable covariance structure
  1460. svd_0 ~~ svd_0
  1461. svd_1 ~~ 0*svd_1
  1462. dC ~~ dC
  1463. dC ~~ svd_0
  1464. ')
  1466. # Fit LCS model with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  1467. fit5 <- lavaan::sem(model5, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  1468. summary(fit5, fit.measures = TRUE, standardized = TRUE)
  1470. # Inspect fit indices
  1471. fitmeasures(fit5, fit.measures = c("chisq.scaled", "pvalue.scaled",
  1472. "cfi.robust", "rmsea.robust", "srmr"))
  1474. # Extract parameter estimates of interest
  1475. par <- parameterEstimates(fit5)
  1476. par[par$lhs == "dC" & grepl("sbp_group_n", par$rhs),]
  1479. #//-------------------------------------------- ANALYSES 3.4: BP MEDIATION OF THE TREATMENT ASSIGNMENT EFFECT --------------------------------------------
  1480. # LCS model with treatment group as latent change score predictor and delta SBP as mediator
  1481. model6 <-paste0('
  1482. r_pvwml.0 ~ age.c.0
  1483. r_pvwml.1 ~ age.c.0 + time_years.1
  1484. r_fw.0 ~ age.c.0
  1485. r_fw.1 ~ age.c.0 + time_years.1
  1486. r_bgepvsc.0 ~ age.c.0
  1487. r_bgepvsc.1 ~ age.c.0 + time_years.1
  1489. # Measurement model for svd in time 0
  1490. svd_0 =~ 1*r_pvwml.0 + fw*r_fw.0 + bg*r_bgepvsc.0
  1492. # Measurement model for svd in time 1
  1493. svd_1 =~ 1*r_pvwml.1 + fw*r_fw.1 + bg*r_bgepvsc.1
  1495. # Mean structure specification
  1496. r_pvwml.0 ~ 0*1
  1497. r_fw.0 ~ ifw*1
  1498. r_bgepvsc.0 ~ ibg*1
  1499. r_pvwml.1 ~ 0*1
  1500. r_fw.1 ~ ifw*1
  1501. r_bgepvsc.1 ~ ibg*1
  1503. # Residual covariances
  1504. r_pvwml.0 ~~ r_pvwml.1
  1505. r_fw.0 ~~ r_fw.1
  1506. r_bgepvsc.0 ~~ r_bgepvsc.1
  1508. # Latent change
  1509. svd_1 ~ 1*svd_0
  1510. dC =~ 1*svd_1
  1512. # Latent variable mean structure
  1513. svd_0 ~ 1
  1514. svd_1 ~ 0*1
  1515. sbp_delta ~ a*treat
  1516. dC ~ 1 + b*sbp_delta + c*treat
  1518. # Latent variable covariance structure
  1519. svd_0 ~~ svd_0
  1520. svd_1 ~~ 0*svd_1
  1521. dC ~~ dC
  1522. dC ~~ svd_0
  1524. Indirect := a*b
  1525. Direct := c
  1526. Total := (a*b) + c
  1527. ')
  1529. # Fit LCS model with Maximum Likelihood estimation method using robust (Huber-White) standard errors and a scaled (Yuan-Bentler) test statistic
  1530. fit6 <- lavaan::sem(model6, data = wide01, estimator = "ML", se = "robust.huber.white", test = "yuan.bentler.mplus")
  1531. summary(fit6, fit.measures = TRUE, standardized = TRUE)
  1533. # Inspect fit indices
  1534. fitmeasures(fit6, fit.measures = c("chisq.scaled", "pvalue.scaled",
  1535. "cfi.robust", "rmsea.robust", "srmr"))
  1537. # Extract parameter estimates of interest
  1538. parameterEstimates(fit6)[parameterEstimates(fit6)$op %in% ":=",
  1539. c("lhs", "est", "ci.lower", "ci.upper")]
  1541. # Path diagram of indirect (through delta SBP) and direct treatment effect (Supplemental Figure 3)
  1542. pl_mod <- semPlotModel(fit6)
  1544. pl_par <- pl_mod@Pars
  1546. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "int" & !pl_par$rhs %in% c("dC", "svd_0"), ]), ]
  1548. pl_par <- pl_par[!rownames(pl_par) %in% rownames(pl_par[pl_par$edge == "<->" & pl_par$rhs %in%
  1549. c("sbp_delta", "treat"), ]), ]
  1551. pl_mod@Pars <- pl_par
  1553. plt <- semPaths(pl_mod, whatLabels = "est", residuals = FALSE, intStyle = "multi", sizeInt = 9, edge.label.cex = 1.2)
  1555. plt$layout[1, ] <- c(-1, 0.6) #p.0
  1556. plt$layout[2, ] <- c(-0.7, 0.6) #f.0
  1557. plt$layout[3, ] <- c(-0.4, 0.6) #b.0
  1558. plt$layout[4, ] <- c(0.2, 0.6) #p.1
  1559. plt$layout[5, ] <- c(0.5, 0.6) #f.1
  1560. plt$layout[6, ] <- c(0.8, 0.6) #b.1
  1562. plt$layout[7, ] <- c(0.8, -0.7) #delta_sbp
  1563. plt$layout[8, ] <- c(0.8, -1.3) #treat
  1565. plt$layout[9, ] <- c(-0.7, 1.35) #age.c.0
  1566. plt$layout[10, ] <- c(0.5, 1.35) #dufu
  1567. plt$layout[11, ] <- c(-0.7, -0.5) #s_0
  1568. plt$layout[12, ] <- c(0.5, -0.5) #s_1
  1569. plt$layout[13, ] <- c(-0.2, -1) #dC
  1570. plt$layout[14, ] <- c(-1.2, -0.5) #int.s_0
  1571. plt$layout[15, ] <- c(-0.8, -1) #int.dC
  1573. plt$graphAttributes$Nodes$labels <- c("Perive-\nticular\nWMH\nt0", "WM\nFree\nWater\nt0", "Basal\nganglia\nPVS\nt0",
  1574. "Perive-\nticular\nWMH\nt1", "WM\nFree\nWater\nt1", "Basal\nganglia\nPVS\nt1",
  1575. "\u0394SBP", "Intensive\nBP\nControl",
  1576. "Base-\nline\nage", "Follow\nup\ntime",
  1577. "SVD\nt0", "SVD\nt1", "SVD\nChange",
  1578. "Mean\nBaseline\nSVD", "Mean\nSVD\nChange")
  1580. plt <- semptools::mark_sig(semPaths_plot = plt, object = fit6)
  1582. plt$graphAttributes$Nodes$label.cex <- 1.6
  1584. plt$graphAttributes$Edges$curve[c(26,27)] <- 0.5
  1586. plt$graphAttributes$Edges$label.margin[c(16:18)] <- -0.035
  1587. plt$graphAttributes$Edges$label.margin[c(21)] <- -0.03
  1588. plt$graphAttributes$Edges$label.margin[c(23)] <- -0.02
  1589. plt$graphAttributes$Edges$label.margin[c(24)] <- -0.02
  1591. plt$plotOptions$label.prop <- c(1, 1, 1,
  1592. 1, 1, 1,
  1593. 0.9, 0.95,
  1594. 0.85, 0.85,
  1595. 0.7, 0.7, 0.85,
  1596. 0.45, 0.425)
  1598. plt$graphAttributes$Edges$color[c(1,2,4,5,7,8)] <- "skyblue"
  1600. plt$graphAttributes$Edges$color[c(3,6,9)] <- "navy"
  1602. plt$graphAttributes$Edges$color[c(10:15)] <- "coral"
  1604. plt$graphAttributes$Edges$color[c(16:18)] <- "lightgreen"
  1606. plt$graphAttributes$Edges$color[c(21,23)] <- "blue"
  1608. plt$graphAttributes$Edges$color[c(22,24)] <- "red2"
  1610. plt$graphAttributes$Nodes$width <- c(5, 5, 5,
  1611. 5, 5, 5,
  1612. 5, 5,
  1613. 5, 5,
  1614. 8, 8, 8,
  1615. 12, 12)
  1617. plot(plt)
  1619. # Bootstrapped solution
  1620. # Detect cpu number
  1621. ncpus = max(1, parallel::detectCores() - 1)
  1623. # Fit LCS model with Maximum Likelihood estimation method using bootstrapped standard errors and a scaled (Yuan-Bentler) test statistic
  1624. fit6_boot <- lavaan::sem(model6, data = wide01, estimator = "ML", test = "yuan.bentler.mplus",
  1625. se = "bootstrap", bootstrap = 5000, parallel = "snow",
  1626. ncpus = ncpus, iseed = 2025)
  1628. # Extract parameter estimates of interest from bootstrapped solution
  1629. sum <- parameterEstimates(fit6_boot)
  1630. sum <- sum[sum$op == ":=" , c("label", "est", "ci.lower", "ci.upper", "pvalue")]
  1631. rownames(sum) <- sum[,"label"]
  1632. sum <- sum[ ,-1]
  1633. sum
  1636. #//--------------------------------------------------------------------------- END OF ANALYTIC CODE ---------------------------------------------------------------------------//

SVD_SPRINT_Git_20260705.R at commit ce6d95d, no license · at the source

Overview

Authors: Sokratis Charisis1,2,3, Nicholas M. Pajewski4, Larry R. Price5, Ngoc Huynh Ho1, Tanweer Rashid1, David Wang1, Yuheng Zeng1, Niyas Shamsudeen Kutty1, R. Nick Bryan6, Kyle C. Kern7, Bradford C. Dickerson2,3, Sudha Seshadri1, Ilya Nasrallah6, Lenore Launer8, Christos Davatzikos6, Jeff D. Williamson9, Mohamad Habes1
ORCID iDs: Sokratis Charisis, Tanweer Rashid, Niyas Shamsudeen Kutty, Kyle C. Kern, Lenore Launer, Mohamad Habes
  1. Neuroimage Analytics Laboratory (NAL) and the Biggs Institute Neuroimaging Core (BINC), Glenn Biggs Institute for Alzheimer's and Neurodegenerative Diseases, University of Texas Health Science Center at San Antonio, San Antonio, TX, USA
  2. Department of Neurology, Massachusetts General Hospital, Boston, MA, USA
  3. Harvard Medical School, Boston, MA, USA
  4. Department of Biostatistics and Data Science, Wake Forest University School of Medicine, Winston-Salem, NC, USA
  5. Department of Psychiatry and Behavioral Sciences, University of Texas Health Science Center at San Antonio, San Antonio, TX, USA
  6. Department of Radiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA
  7. Department of Neurology, UCLA David Geffen School of Medicine, Los Angeles, CA, USA
  8. Neuroepidemiology Section, Intramural Research Program, National Institute on Aging, Bethesda, MD, USA
  9. Section of Gerontology and Geriatric Medicine, Department of Internal Medicine, Wake Forest University School of Medicine, Winston-Salem, NC, USA
Institutions: Harvard University (United States); The University of Texas at San Antonio Health Science Center (United States); Massachusetts General Hospital (United States); Wake Forest University Health Sciences (United States); Wake Forest University (United States); University of Pennsylvania (United States); University of California, Los Angeles (United States); National Institutes of Health (United States); National Institute on Aging (United States)
Journal: EClinicalMedicine, volume 99, article 104143
Dates: received 15 February 2026; accepted 24 July 2026; published online 11 August 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.1016/j.eclinm.2026.104143 · PMID 42614618 · PMCID PMC13482511 · OpenAlex W7202205841
Open access: gold, a free copy (OpenAlex)
Status: code verified
Categories: structural MRI / diffusion (modality), human (organism), stroke (population), clinical / translational (subfield)
Methods: Statistics, fMRI & imaging
Keywords: Cerebral small vessel disease, Intensive blood pressure treatment, Dose–response relationship, Randomized clinical trial, Brain MRI, Neuroimaging biomarkers, SPRINT
Topic: Cerebrospinal fluid and hydrocephalus (Cellular and Molecular Neuroscience, Neuroscience), according to OpenAlex
Funding: National Institutes of Health; NIH; HHS | NIH | National Heart, Lung, and Blood Institute (NHLBI); NIDDK; National Institute on Aging; NINDS; Kulynych Family Foundation; Oristano Family Foundation; National Center for Advancing Translational Sciences; Case Western Reserve University; Ohio State University; University of Pennsylvania; Stanford; Tufts Medical Center; University of Illinois at Urbana-Champaign; University of Pittsburgh; University of Texas Southwestern Medical Center; University of Utah; Vanderbilt University; George Washington University; University of California, Davis; University of Florida; University of Michigan; Wake Forest Claude Pepper Center; Wake Forest University Alzheimer’s Disease Research Center
Citations: not cited yet (Europe PMC); 38 references in the paper

Abstract

Background: The benefits of improved systolic blood pressure (SBP) control on stroke, coronary heart disease, and heart failure are well-established, yet its effect on overall cerebral small vessel disease (SVD) burden remains uncharacterized. We examined the association between intensive SBP control and change in SVD burden.

Methods: We conducted a post-hoc analysis of the Systolic Blood Pressure Intervention Trial (SPRINT), a multicenter randomized clinical trial. Of 1267 hypertensive individuals aged ≥50 years without diabetes or prior stroke screened for the brain MRI substudy, 663 and 442 participants completed brain MRI that met quality control criteria and had complete data on SVD indicators at baseline and at a median of 3.9 (interquartile range, 3.6–4.1) years after randomization, respectively. From November 2010 to March 2013, participants were randomly assigned to an intensive SBP target of <120 mmHg (n = 348) or a standard target of <140 mmHg (n = 315). Post-hoc outcome was change in a global SVD factor, longitudinally validated using confirmatory factor analysis and designed to capture overall SVD-related vascular brain injury by integrating three complementary imaging endophenotypes: periventricular white matter hyperintensities, white matter free water, and basal ganglia perivascular spaces. This trial is registered with ClinicalTrials.gov (NCT01206062).

Findings: Mean [SD] baseline age was 68.1 (8.6) years; 263 [40%] participants were women. Compared with standard SBP treatment, intensive treatment was associated with significantly less SVD progression (standardized mean difference [Cohen's d] = −0.40 [95% CI, −0.62 to −0.17]). We also observed gradually more favorable SVD burden changes with greater attained SBP reductions, demonstrating a clear dose–response relationship: 21.2% (95% CI, 7.4%–35%), 26.3% (13.1%–39.5%), and 39.4% (24.2%–54.5%) less progression relative to baseline SVD burden for SBP reductions of 0–10 mmHg, 10–20 mmHg, and ≥20 mmHg, respectively.

Interpretation: Among hypertensive adults, targeting an SBP of <120 mmHg, compared with <140 mmHg, was associated with less progression of SVD burden. Even modest SBP reductions of ≤10 mmHg conferred measurable brain benefits, with larger reductions providing incrementally greater protection against SVD progression.

Funding: National Institutes of Health, National Heart, Lung, and Blood Institute, National Institute of Diabetes and Digestive and Kidney Diseases, National Institute on Aging, and National Institute of Neurological Disorders and Stroke.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repository

Its files are read in the Code ↔ Paper reader above, with 7 matches between paragraphs and lines of code.

UTHSCSA-NAL/SPRINT_SVD

License: none: the authors keep all their rights
State: the link answers, verified on 27 September 2026
Evidence: files inventoried
Commit: ce6d95d69bd81a9f15cb1cffdd7011a7fdb14290, 5 July 2026
Languages: R (1)
Size: 2 files, 1 script
Software Heritage: not archived
Found in: “Data sharing statement”
Holds: README
Not found: license file, CITATION.cff, environment file, tests, continuous integration, documentation
Tools: lavaan (1 file), tidyverse (1 file)
Availability: 1 check, the latest on 27 September 2026: the link answers
  • 27 September 2026: the link answers
2 files
At the source: github.com/UTHSCSA-NAL/SPRINT_SVD

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 1 script, each with its path and the digest of its content;
  • 7 matches between paragraphs of the paper and lines of the code (method lexical-v1);
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Data sharing statement

Deidentified participant data, along with relevant data dictionaries, are available through BioLINCC (https://biolincc.nhlbi.nih.gov/studies/sprint) to investigators who provide an Institutional Review Board/Ethics approval, or certification of exemption from Institutional Review Board/Ethics review, and who agree to the terms and conditions of a data use agreement. Data are available for any purpose, unless prohibited by the informed consent. The source code for the conducted analyses is available at https://github.com/UTHSCSA-NAL/SPRINT_SVD.

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 3, 28 September 2026

  • Authors: added Tanweer Rashid (0000-0001-6844-9558); Niyas Shamsudeen Kutty (0000-0001-5511-4249); Kyle C. Kern (0000-0002-2703-7669); Lenore Launer (0000-0002-3238-7612); Mohamad Habes (0000-0001-9447-5805); removed Tanweer Rashid; Niyas Shamsudeen Kutty; Kyle C. Kern; Lenore Launer; Mohamad Habes

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, pages, dates, 17 authors, 7 keywords, 25 funders, 36 references.

Cite

This paper

Charisis, S., Pajewski, N. M., Price, L. R., Ho, N. H., Rashid, T., Wang, D., Zeng, Y., Kutty, N. S., Bryan, R. N., Kern, K. C., Dickerson, B. C., Seshadri, S., Nasrallah, I., Launer, L., Davatzikos, C., Williamson, J. D., & Habes, M. (2026). Intensive versus standard blood pressure control and overall brain small vessel disease burden: a post-hoc analysis of the SPRINT randomized clinical trial. EClinicalMedicine, 99, 104143. https://doi.org/10.1016/j.eclinm.2026.104143

BibTeX

@article{charisis2026intensive,
author = {Charisis, Sokratis and Pajewski, Nicholas M. and Price, Larry R. and Ho, Ngoc Huynh and Rashid, Tanweer and Wang, David and Zeng, Yuheng and Kutty, Niyas Shamsudeen and Bryan, R. Nick and Kern, Kyle C. and Dickerson, Bradford C. and Seshadri, Sudha and Nasrallah, Ilya and Launer, Lenore and Davatzikos, Christos and Williamson, Jeff D. and Habes, Mohamad},
title = {{Intensive versus standard blood pressure control and overall brain small vessel disease burden: a post-hoc analysis of the SPRINT randomized clinical trial}},
journal = {EClinicalMedicine},
year = {2026},
month = aug,
volume = {99},
pages = {104143},
publisher = {Elsevier},
issn = {2589-5370},
doi = {10.1016/j.eclinm.2026.104143},
url = {https://doi.org/10.1016/j.eclinm.2026.104143},
pmid = {42614618},
pmcid = {PMC13482511}
}

RIS

TY - JOUR
AU - Charisis, Sokratis
AU - Pajewski, Nicholas M.
AU - Price, Larry R.
AU - Ho, Ngoc Huynh
AU - Rashid, Tanweer
AU - Wang, David
AU - Zeng, Yuheng
AU - Kutty, Niyas Shamsudeen
AU - Bryan, R. Nick
AU - Kern, Kyle C.
AU - Dickerson, Bradford C.
AU - Seshadri, Sudha
AU - Nasrallah, Ilya
AU - Launer, Lenore
AU - Davatzikos, Christos
AU - Williamson, Jeff D.
AU - Habes, Mohamad
TI - Intensive versus standard blood pressure control and overall brain small vessel disease burden: a post-hoc analysis of the SPRINT randomized clinical trial
T2 - EClinicalMedicine
J2 - eClinicalMedicine
PY - 2026
DA - 2026/08/11
VL - 99
SP - 104143
SN - 2589-5370
PB - Elsevier
DO - 10.1016/j.eclinm.2026.104143
UR - https://doi.org/10.1016/j.eclinm.2026.104143
LA - en
ER -

CSL-JSON

{
"id": "10.1016/j.eclinm.2026.104143",
"type": "article-journal",
"title": "Intensive versus standard blood pressure control and overall brain small vessel disease burden: a post-hoc analysis of the SPRINT randomized clinical trial",
"container-title": "EClinicalMedicine",
"author": [
{
"family": "Charisis",
"given": "Sokratis"
},
{
"family": "Pajewski",
"given": "Nicholas M."
},
{
"family": "Price",
"given": "Larry R."
},
{
"family": "Ho",
"given": "Ngoc Huynh"
},
{
"family": "Rashid",
"given": "Tanweer"
},
{
"family": "Wang",
"given": "David"
},
{
"family": "Zeng",
"given": "Yuheng"
},
{
"family": "Kutty",
"given": "Niyas Shamsudeen"
},
{
"family": "Bryan",
"given": "R. Nick"
},
{
"family": "Kern",
"given": "Kyle C."
},
{
"family": "Dickerson",
"given": "Bradford C."
},
{
"family": "Seshadri",
"given": "Sudha"
},
{
"family": "Nasrallah",
"given": "Ilya"
},
{
"family": "Launer",
"given": "Lenore"
},
{
"family": "Davatzikos",
"given": "Christos"
},
{
"family": "Williamson",
"given": "Jeff D."
},
{
"family": "Habes",
"given": "Mohamad"
}
],
"container-title-short": "eClinicalMedicine",
"volume": "99",
"page": "104143",
"DOI": "10.1016/j.eclinm.2026.104143",
"PMID": "42614618",
"PMCID": "PMC13482511",
"ISSN": "2589-5370",
"publisher": "Elsevier",
"URL": "https://doi.org/10.1016/j.eclinm.2026.104143",
"language": "en",
"issued": {
"date-parts": [
[
2026,
8,
11
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1038/s41467-026-74515-w [code]
Advancing fair and explainable machine learning for neuroimaging dementia pattern classification in multi-racial and multi-ethnic populations.
Journal: Nature communications
In common: structural MRI / diffusion, 2 references, 2 authors
[2] doi:10.3389/fneur.2026.1879894
Perivascular diffusion and white matter free water improve identification of cognitive impairment in cerebral small vessel disease.
Journal: Frontiers in neurology
In common: stroke, structural MRI / diffusion, clinical / translational, 4 references
[3] doi:10.1093/braincomms/fcag045 [code]
Examining neuroimaging biomarkers, plasma biomarkers and cognitive functions in patients with recovered COVID-19 infection: a multicentre study using 7T MRI.
Journal: Brain communications
In common: stroke, structural MRI / diffusion, clinical / translational, author Mohamad Habes
[4] doi:10.1038/s41467-026-72091-7 [code]
Coupled cross-sectional and longitudinal non-negative matrix factorization reveals dominant brain aging trajectories in 48,949 individuals.
Journal: Nature communications
In common: structural MRI / diffusion, 4 references
[5] doi:10.1212/wnl.0000000000218472 [code]
Lesion-Level Subtypes of White Matter Hyperintensity Evolution Beyond Spatial Location.
Journal: Neurology
In common: stroke, structural MRI / diffusion, clinical / translational, 3 references
[6] doi:10.1016/j.nicl.2026.104001 [code]
Effect of vascular lesion preprocessing on Brain Intensity AbNormality Classification Algorithm (BIANCA) white matter hyperintensity segmentation.
Journal: NeuroImage. Clinical
In common: stroke, structural MRI / diffusion, 3 references
[7] doi:10.64898/2026.03.09.26347914 [code]
Multimodal Ageing Biomarkers and Plasma Proteomic Signatures Associated with All-Cause Mortality
Journal: medRxiv (preprint)
In common: lavaan, tidyverse, structural MRI / diffusion, clinical / translational, 1 reference
[8] doi:10.1093/braincomms/fcag080 [code]
Synergistic effects of cerebral small vessel disease burden and plasma phosphorylated tau 181 on white matter microstructure and cognition in a Chinese cohort.
Journal: Brain communications
In common: tidyverse, stroke, structural MRI / diffusion, 2 references
[9] doi:10.1212/wnl.0000000000218164 [code]
Race and Ethnicity, Hypertension, and Neuroimaging Markers of Brain Aging: A Causal Mediation Analysis in the HABS-HD Study.
Journal: Neurology
In common: tidyverse, structural MRI / diffusion, clinical / translational, 2 references
[10] doi:10.3389/fnut.2026.1837406 [code]
MIND diet moderates the associations between cerebrovascular and neurodegenerative disease burden and cognition.
Journal: Frontiers in nutrition
In common: tidyverse, stroke, structural MRI / diffusion, 2 references

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.