library(shiny)
library(bslib)
library(mrgsolve)
library(dplyr)
library(ggplot2)
library(tidyr)

# ---- mrgsolve model ----
# Unit system: T-DXd amounts in mg, DXd amounts in µg, volumes in L, time in hours
# CP_TDXd in ng/mL (mg/L × 1000), CP_DXd in ng/mL (µg/L = ng/mL)
model_code <- '
$PARAM @annotated
CLintact : 0.01754 : T-DXd clearance base (L/h, = 0.421 L/day)
V1       : 2.77   : T-DXd central volume (L)
Qintact  : 0.2146 : T-DXd distributional CL (L/h, = 5.15 L/day)
V2       : 5.16   : T-DXd peripheral volume (L)
CLdrug   : 19.2   : DXd clearance base (L/h)
KREL_C1  : 0.0159 : Release-rate constant cycle 1 (1/h)
WT       : 57.8   : Body weight (kg)
ALB      : 40.0   : Albumin (g/L)
TSIZE    : 57.0   : Baseline tumor size (mm)
SEX      : 0      : Sex (0=female, 1=male)
JAPAN    : 0      : Japanese (0=no, 1=yes)
BILI     : 8.0    : Total bilirubin (umol/L)
AST      : 30.0   : AST (U/L)

$CMT @annotated
CENT1 : T-DXd central amount (mg)
PERI1 : T-DXd peripheral amount (mg)
DXD   : Released DXd amount (ug)

$MAIN
// Covariate-adjusted T-DXd parameters
double CLi = CLintact
             * pow(WT/57.8, 0.370)
             * pow(ALB/40.0, -0.533)
             * pow(TSIZE/57.0, 0.071)
             * (JAPAN == 1 ? 0.903 : 1.0)
             * (SEX == 1 ? 1.174 : 1.0);

double V1i = V1
             * pow(WT/57.8, 0.489)
             * (SEX == 1 ? 1.197 : 1.0);

double V2i = V2 * (JAPAN == 1 ? 0.738 : 1.0);

// Covariate-adjusted DXd parameters
double CLd = CLdrug
             * pow(WT/57.8, 0.440)
             * pow(BILI/8.0, -0.207)
             * pow(AST/30.0, -0.219);

// Body surface area (DuBois, HT fixed 163 cm for typical female)
double BSA_m2 = 0.007184 * pow(WT, 0.425) * pow(163.0, 0.725);
double Vd = 17.0 * BSA_m2;

$ODE
double C1 = CENT1 / V1i;   // mg/L
double C2 = PERI1 / V2i;   // mg/L

// Cycle-dependent release-rate constant
int cyc = (int)(SOLVERTIME / 504.0) + 1;
if (cyc < 1) cyc = 1;
double KREL = KREL_C1 * pow((double)cyc, -0.137);
if (cyc > 1) KREL = KREL * 0.830;

// T-DXd 2-CMT ODEs (amounts in mg)
dxdt_CENT1 = -CLi * C1 - Qintact * C1 + Qintact * C2;
dxdt_PERI1 =  Qintact * C1 - Qintact * C2;

// DXd 1-CMT ODE (amount in µg)
// DAR*MW_DXd/MW_TDXd*1000 = 8*600/156000*1000 = 30.77 µg DXd per mg T-DXd
dxdt_DXD = KREL * CENT1 * 30.77 - (CLd / Vd) * DXD;

$TABLE
double CP_TDXd = (CENT1 / V1i) * 1000.0;   // ng/mL (mg/L → ng/mL)
double CP_DXd  = DXD / Vd;                  // ng/mL (µg/L = ng/mL)

$CAPTURE CP_TDXd CP_DXd
'

mod <- mcode("tdxd_popPK", model_code, quiet = TRUE)

# ---- Theme ----
app_theme <- bs_theme(
  version = 5, bootswatch = "flatly", primary = "#8b5cf6"
) |> bs_add_rules("
  .metric-card {
    background: #f8f9fa; border-radius: 8px; padding: 15px; margin: 5px;
    text-align: center; border: 1px solid #dee2e6;
  }
  .metric-value { font-size: 22px; font-weight: bold; color: #2c3e50; }
  .metric-label { font-size: 11px; color: #7f8c8d; margin-top: 3px; }
  .metric-success .metric-value { color: #10b981; }
  .metric-warning .metric-value { color: #f59e0b; }
  .metric-primary .metric-value { color: #8b5cf6; }
  .metric-info .metric-value    { color: #0dcaf0; }
  .ref-box {
    background: #f0f4ff; border-left: 4px solid #8b5cf6;
    padding: 12px 16px; border-radius: 4px; margin-top: 10px; font-size: 13px;
  }
  .ref-box a { color: #8b5cf6; }
  .section-label { font-size: 11px; font-weight: 600; color: #6c757d;
    text-transform: uppercase; letter-spacing: 0.05em; margin-bottom: 4px; }
")

# ---- UI ----
ui <- page_sidebar(
  title = "Trastuzumab Deruxtecan (T-DXd / ENHERTU) PK Simulator",
  theme = app_theme,

  sidebar = sidebar(
    title = "Simulation Settings", width = 330,

    div(class = "section-label", "Dosing"),
    sliderInput("dose_mgkg", "Dose (mg/kg)", min = 0.8, max = 8.0,
                value = 5.4, step = 0.1),
    tags$small(class = "text-muted",
      "Approved dose: 5.4 mg/kg Q3W IV. Infusion 90 min."),
    br(), br(),
    sliderInput("n_cycles", "Number of Cycles", min = 1, max = 12,
                value = 6, step = 1),

    hr(),
    div(class = "section-label", "Patient Characteristics"),
    sliderInput("wt", "Body Weight (kg)", min = 35, max = 130,
                value = 58, step = 1),
    sliderInput("alb", "Albumin (g/L)", min = 25, max = 50,
                value = 40, step = 1),
    sliderInput("tsize", "Baseline Tumor Size (mm)", min = 10, max = 200,
                value = 57, step = 5),
    selectInput("sex", "Sex", choices = c("Female", "Male"), selected = "Female"),
    checkboxInput("japan", "Japanese patient", value = FALSE),

    hr(),
    div(class = "section-label", "Hepatic Markers (DXd CL)"),
    sliderInput("bili", "Total Bilirubin (µmol/L)", min = 2, max = 30,
                value = 8, step = 1),
    sliderInput("ast", "AST (U/L)", min = 10, max = 120,
                value = 30, step = 5),

    hr(),
    checkboxInput("log_scale", "Log Scale (Y-axis)", value = TRUE)
  ),

  navset_card_underline(
    title = "Trastuzumab Deruxtecan (T-DXd) PK Simulator",
    full_screen = TRUE,

    nav_panel("Simulation",
      layout_column_wrap(
        width = 1/4, fill = FALSE,
        div(class = "metric-card metric-success",
          div(class = "metric-value", textOutput("tdxd_cmax")),
          div(class = "metric-label", "T-DXd Cmax (ng/mL)")
        ),
        div(class = "metric-card metric-warning",
          div(class = "metric-value", textOutput("tdxd_ctrough")),
          div(class = "metric-label", "T-DXd Ctrough (ng/mL)")
        ),
        div(class = "metric-card metric-primary",
          div(class = "metric-value", textOutput("tdxd_thalf")),
          div(class = "metric-label", "T-DXd t\u00bd (days)")
        ),
        div(class = "metric-card metric-info",
          div(class = "metric-value", textOutput("dxd_cmax")),
          div(class = "metric-label", "DXd Cmax (ng/mL)")
        )
      ),
      card(
        full_screen = TRUE, height = "500px",
        plotOutput("pkPlot", height = "100%")
      )
    ),

    nav_panel("Model Information",
      div(class = "ref-box",
        markdown("
## Trastuzumab Deruxtecan — Sequential Population PK Model

**Drug:** Trastuzumab deruxtecan (T-DXd, DS-8201, ENHERTU) — HER2-targeting ADC  
**Indication:** HER2-positive/HER2-low/HER2-ultralow unresectable or metastatic breast cancer, HER2-mutant NSCLC  
**Route:** IV infusion Q3W  
**Approved dose:** 5.4 mg/kg Q3W (breast cancer); 6.4 mg/kg Q3W for some indications

---

### Intact T-DXd (ADC) — 2-Compartment IV Model

| Parameter | Typical Value | Units |
|-----------|--------------|-------|
| CL (CLintact) | 0.421 | L/day |
| V1 (central) | 2.77 | L |
| Q (distributional CL) | 5.15 | L/day |
| V2 (peripheral) | 5.16 | L |
| Terminal t½ | ~13.5 | days |
| BSV (CL) | 25% | %CV |
| BSV (V1) | 16% | %CV |

**Covariate effects on CLintact:**  
`CL = 0.421 × (WT/57.8)^0.370 × (ALB/40)^-0.533 × (TSize/57)^0.071 × [×0.903 if Japanese] × [×1.174 if male]`

**Covariate effects on V1:**  
`V1 = 2.77 × (WT/57.8)^0.489 × [×1.197 if male]`

---

### Released Drug (DXd) — 1-Compartment Formation-Rate Limited

| Parameter | Typical Value | Units |
|-----------|--------------|-------|
| CL (CLdrug) | 19.2 | L/h |
| V (Vdrug) | 17 × BSA | L |
| Krel (Cycle 1) | 0.0159 | 1/h |
| Intrinsic t½ | ~1 h | — |
| Apparent t½ | ~13.5 days | formation-rate limited |

**Time-varying Krel:**  
`Krel = 0.0159 × Cycle^(-0.137) × [×0.830 if Cycle > 1]`  
(39% reduction at Cycle 10; ~45% at Cycle 20 vs Cycle 1)

**Covariate effects on CLdrug:**  
`CL = 19.2 × (WT/57.8)^0.440 × (BILI/8)^-0.207 × (AST/30)^-0.219`

---

### Key Findings
- All covariates (WT, albumin, tumor size, sex, race) produced <20% change in AUC at steady state, except extreme body weights (>86 kg: +28% AUC) and very low albumin (<31 g/L: -23% Cmin)
- **No dose adjustment required** based on tested covariates, race, or mild/moderate organ impairment
- DXd PK is formation-rate limited — apparent t½ driven by T-DXd elimination

### DAR/MW Scaling
DXd release modeled with MW-based scaling: 8 × (600 g/mol DXd)/(156,000 g/mol T-DXd) × 1000 = 30.77 µg DXd per mg T-DXd
        ")
      )
    ),

    nav_panel("References",
      div(class = "ref-box",
        tags$h5("\U0001f4da Key References"),
        tags$ol(
          tags$li(
            "Yin O, et al. ",
            tags$em("Population Pharmacokinetics of Trastuzumab Deruxtecan in Patients With HER2-Positive Breast Cancer and Other Solid Tumors."),
            " Clin Pharmacol Ther. 2021;109(5):1314-1325. ",
            tags$a(href = "https://doi.org/10.1002/cpt.2096", target = "_blank", "doi:10.1002/cpt.2096")
          ),
          tags$li(
            "Modi S, et al. Trastuzumab Deruxtecan in Previously Treated HER2-Positive Breast Cancer. ",
            tags$em("N Engl J Med."), " 2020;382:610–621."
          ),
          tags$li(
            "FDA Label: ENHERTU (fam-trastuzumab deruxtecan-nxki). ",
            tags$a(href = "https://www.accessdata.fda.gov/drugsatfda_docs/label/2025/761139s032s035lbl.pdf",
                   target = "_blank", "Prescribing Information (2025)")
          )
        ),
        tags$h5("\U0001f48a Drug Context"),
        tags$ul(
          tags$li(tags$strong("Class:"), " Antibody-Drug Conjugate (ADC), HER2-targeting"),
          tags$li(tags$strong("Mechanism:"), " Anti-HER2 mAb + cleavable linker + topoisomerase I inhibitor (exatecan derivative, DAR ≈ 8)"),
          tags$li(tags$strong("Approved indications:"),
            " HER2+ mBC, HER2-low/ultralow mBC, HER2-mutant NSCLC, HER2+ gastric/GEJ cancer"),
          tags$li(tags$strong("Approved doses:"),
            " 5.4 mg/kg Q3W (HER2+ BC, HER2-low/ultralow BC, NSCLC); 6.4 mg/kg Q3W (HER2+ gastric)"),
          tags$li(tags$strong("Dataset:"),
            " 639 patients, 5 clinical studies (DESTINY-Breast01 + 4 phase I studies), doses 0.8–8.0 mg/kg Q3W")
        )
      )
    )
  ),

  div(
    style = paste0("text-align: center; padding: 20px; margin-top: 30px;",
                   " border-top: 1px solid #e9ecef; color: #6c757d; font-size: 12px;"),
    "Powered by ",
    tags$a(href = "https://www.pkpdbuilder.com", target = "_blank",
           style = "color: #8b5cf6; font-weight: 500;", "PKPDBuilder.com"),
    " \u2022 Yin et al. Clin Pharmacol Ther 2021 \u2022 For research and educational purposes only",
    br(),
    tags$span(style = "font-size: 10px;",
      "Not for clinical decision-making. Population-typical values shown.")
  )
)

# ---- Server ----
server <- function(input, output, session) {

  # ---- Simulation ----
  sim_data <- reactive({
    shiny::req(input$dose_mgkg, input$wt, input$n_cycles)

    wt      <- input$wt
    dose_mg <- input$dose_mgkg * wt
    nc      <- input$n_cycles
    sex_val <- ifelse(input$sex == "Male", 1, 0)
    jp_val  <- ifelse(isTRUE(input$japan), 1, 0)

    # IV infusion: 90 min = 1.5 h rate
    infusion_h <- 1.5
    rate_mgh   <- dose_mg / infusion_h

    # Dosing events: one per cycle (Q3W = 504 h)
    ev_list <- lapply(seq_len(nc) - 1, function(c_idx) {
      ev(amt = dose_mg, cmt = 1,
         time = c_idx * 504,
         rate = rate_mgh)
    })
    ev_all <- do.call(c, ev_list)

    # Simulation end: end of last cycle + 3 days buffer
    end_h <- nc * 504 + 72

    mod |>
      param(WT     = wt,
            ALB    = input$alb,
            TSIZE  = input$tsize,
            SEX    = sex_val,
            JAPAN  = jp_val,
            BILI   = input$bili,
            AST    = input$ast) |>
      ev(ev_all) |>
      mrgsim(end = end_h, delta = 2) |>
      as.data.frame() |>
      dplyr::filter(time >= 0)
  })

  # ---- Helper: last-cycle data ----
  last_cycle_data <- reactive({
    d  <- sim_data()
    nc <- input$n_cycles
    t_start <- (nc - 1) * 504
    dplyr::filter(d, time >= t_start, time <= t_start + 504)
  })

  # ---- Metric: T-DXd Cmax ----
  output$tdxd_cmax <- renderText({
    lc <- last_cycle_data()
    cmax <- max(lc$CP_TDXd, na.rm = TRUE)
    if (cmax >= 1000) {
      sprintf("%.1f k", cmax / 1000)
    } else {
      sprintf("%.0f", cmax)
    }
  })

  # ---- Metric: T-DXd Ctrough ----
  output$tdxd_ctrough <- renderText({
    lc  <- last_cycle_data()
    nc  <- input$n_cycles
    # Trough = last value before next dose (or end of cycle)
    ct <- dplyr::filter(lc, time >= (nc - 1) * 504 + 480) # last ~24h of cycle
    val <- min(ct$CP_TDXd, na.rm = TRUE)
    if (is.finite(val) && val >= 1000) {
      sprintf("%.1f k", val / 1000)
    } else if (is.finite(val)) {
      sprintf("%.0f", val)
    } else {
      "—"
    }
  })

  # ---- Metric: T-DXd t½ ----
  output$tdxd_thalf <- renderText({
    wt     <- input$wt
    alb    <- input$alb
    tsize  <- input$tsize
    sex_v  <- ifelse(input$sex == "Male", 1, 0)
    jp_v   <- ifelse(isTRUE(input$japan), 1, 0)

    CLi <- 0.01754 *
           (wt / 57.8)^0.370 *
           (alb / 40.0)^(-0.533) *
           (tsize / 57.0)^0.071 *
           (if (jp_v == 1) 0.903 else 1.0) *
           (if (sex_v == 1) 1.174 else 1.0)
    V1i <- 2.77 * (wt / 57.8)^0.489 * (if (sex_v == 1) 1.197 else 1.0)
    V2i <- 5.16 * (if (jp_v == 1) 0.738 else 1.0)
    Q   <- 0.2146

    k10 <- CLi / V1i
    k12 <- Q   / V1i
    k21 <- Q   / V2i
    a_sum <- k10 + k12 + k21
    disc  <- sqrt(a_sum^2 - 4 * k10 * k21)
    beta  <- (a_sum - disc) / 2   # h^-1
    thalf_h <- log(2) / beta
    thalf_d <- thalf_h / 24
    sprintf("%.1f", thalf_d)
  })

  # ---- Metric: DXd Cmax ----
  output$dxd_cmax <- renderText({
    lc <- last_cycle_data()
    val <- max(lc$CP_DXd, na.rm = TRUE)
    sprintf("%.2f", val)
  })

  # ---- PK Plot ----
  output$pkPlot <- renderPlot({
    d  <- sim_data()
    nc <- input$n_cycles

    # Reshape to long format
    d_long <- d |>
      dplyr::select(time, CP_TDXd, CP_DXd) |>
      tidyr::pivot_longer(
        cols      = c(CP_TDXd, CP_DXd),
        names_to  = "analyte",
        values_to = "conc"
      ) |>
      dplyr::mutate(
        analyte = dplyr::case_when(
          analyte == "CP_TDXd" ~ "Intact T-DXd (ng/mL)",
          analyte == "CP_DXd"  ~ "Released DXd (ng/mL)"
        ),
        time_day = time / 24
      ) |>
      dplyr::filter(conc > 0)

    # Vertical lines at each dose
    dose_days <- (seq_len(nc) - 1) * 21

    p <- ggplot(d_long, aes(x = time_day, y = conc, color = analyte)) +
      geom_vline(xintercept = dose_days,
                 linetype = "dashed", color = "#e9ecef", linewidth = 0.4) +
      geom_line(linewidth = 0.85) +
      scale_color_manual(
        values = c(
          "Intact T-DXd (ng/mL)" = "#8b5cf6",
          "Released DXd (ng/mL)" = "#f59e0b"
        ),
        name = NULL
      ) +
      facet_wrap(~ analyte, ncol = 1, scales = "free_y") +
      labs(
        x     = "Time (days)",
        y     = "Concentration (ng/mL)",
        title = sprintf(
          "T-DXd %.1f mg/kg Q3W \u00d7 %d cycles (%.0f kg, ALB %.0f g/L%s%s)",
          input$dose_mgkg, nc, input$wt, input$alb,
          if (input$sex == "Male") ", Male" else "",
          if (isTRUE(input$japan)) ", Japanese" else ""
        )
      ) +
      theme_minimal(base_size = 13) +
      theme(
        legend.position  = "none",
        strip.text       = element_text(face = "bold", size = 12),
        strip.background = element_rect(fill = "#f0f4ff", color = NA),
        panel.grid.minor = element_blank(),
        plot.title       = element_text(size = 12, color = "#495057")
      )

    if (input$log_scale) {
      p <- p + scale_y_log10(
        labels = scales::label_number(scale_cut = scales::cut_short_scale())
      )
    } else {
      p <- p + scale_y_continuous(
        labels = scales::label_number(scale_cut = scales::cut_short_scale())
      )
    }

    p
  })
}

shinyApp(ui = ui, server = server)