ICARUS
Immune Checkpoint Antagonism to RedUce Squamous cell carcinoma

Survey Questions

REDCap Survey

https://redcap.link/k77y3o9w

Role

CSCC Patient Care Experience

CSCC Patient Care Experience

CSCC Patient Care Experience

Overview

  • Hypothesis
  • Rationale
  • Schema
    • Schema rationale
  • Efficacy Assessment
    • Statistical analysis
  • Sample Size Calculations
  • Eligibility Criteria
  • Treatment Considerations
  • Alternative Approach

Hypothesis

  • Anti-PD1 therapy will decrease the development of CSCC in high-risk patients with a history of multiple NMSCs

Rationale

  • Effective preventive strategies to mitigate CSCC are lacking
  • Anti-PD1 therapy has shown to be effective in the neoadjuvant and advanced setting
  • Intervention earlier in the disease journey will improve outcomes for patients at high-risk for morbid disease

Schema - Rationale

  • Feasibility
    • Patients with ≥10 NMSCS (≥5 CSCCs) enriches for patients at high risk for recurrent disease
    • Active surveillance for high-risk patients involves q3-6 month TSE (minimizes extra visits)
  • Bias
    • Placebo
      • Important to reduce bias when outcome is determined by physical exam & not scans
  • Efficacy Assessment
    • Primary Endpoint: Rate of new KCC is an established endpoint (e.g.ONTRAC & ONTRANS (NEJM 2015/2023))

Efficacy Assessment

  • Primary endpoint
    • Rate of new CSCC
      • count data
  • Modeling count variables
    • Poisson regression
    • Negative binomial regression
      • Similar structure as Poisson, but has an extra parameter to model over-dispersion
        • When the observed variance is higher than the variance of the theoretical model
          • Common in real-world data sets are populations are often heterogeneous and non-uniform

Efficacy Assessment

library(MASS)

glm.nb(
  formula = `New CSCCs` ~ Treatment + `Baseline CSCCs` + `CLL` + `Previous Radiation`,
  init.theta = 0.75
)

Secondary Endpoints

  • Number of new basal-cell carcinomas & actinic keratoses during the 12-month intervention period
  • The number of new non-melanoma skin cancers in the 6-month post intervention period
  • Safety of anti-PDX as assessed by the numbers and types of adverse events

Sample Size Calculations

  • Considerations
    • Sample size will vary depending on:
      • baseline event rate
      • effect size

Sample Size Calculations

  • Example #1
n <- ynegbinomsize(
  r0=7.0, # event rate group 1 (Placebo)
  r1=4.9, # event rate group 2 (Treatment with 30% rate reduction)
  shape1=0.75, pi1=0.5, alpha=0.05, twosided=1, beta=0.2, 
  fixedfu=1, type=4, u=c(0.5,0.5,1), ut=c(0.5,1.0,1.5), 
  tfix=5, maxfu=1, tchange=1,ratec1=.1, ratec0=.1, eps=1.0e-03)

Total Sample Size: 254

Calculating sample size based on using a Negative Binomial Regression Model (using NBDesign Package).

Sample Size Calculations

  • Example #2
n <- ynegbinomsize(
  r0=2.6, # event rate group 1 (Placebo)
  r1=1.8, # event rate group 2 (Treatment with 30% rate reduction)
  shape1=0.75, pi1=0.5, alpha=0.05, twosided=1, beta=0.2, 
  fixedfu=1, type=4, u=c(0.5,0.5,1), ut=c(0.5,1.0,1.5), 
  tfix=5, maxfu=1, tchange=1,ratec1=.1, ratec0=.1, eps=1.0e-03)

Total Sample Size: 299

Calculating sample size based on using a Negative Binomial Regression Model (using NBDesign Package).

Sample Size Calculations

  • Example #3
n <- ynegbinomsize(
  r0=2.0, # event rate group 1 (Placebo)
  r1=1.4, # event rate group 2 (Treatment with 30% rate reduction)
  shape1=0.75, pi1=0.5, alpha=0.05, twosided=1, beta=0.2, 
  fixedfu=1, type=4, u=c(0.5,0.5,1), ut=c(0.5,1.0,1.5), 
  tfix=5, maxfu=1, tchange=1,ratec1=.1, ratec0=.1, eps=1.0e-03)

Total Sample Size: 348

Calculating sample size based on using a Negative Binomial Regression Model (using NBDesign Package).

Sample Size Calculations

  • Example #4
n <- ynegbinomsize(
  r0=2.0, # event rate group 1 (Placebo)
  r1=1.2, # event rate group 2 (Treatment with 40% rate reduction)
  shape1=0.75, pi1=0.5, alpha=0.05, twosided=1, beta=0.2, 
  fixedfu=1, type=4, u=c(0.5,0.5,1), ut=c(0.5,1.0,1.5), 
  tfix=5, maxfu=1, tchange=1,ratec1=.1, ratec0=.1, eps=1.0e-03)

Total Sample Size: 172

Calculating sample size based on using a Negative Binomial Regression Model (using NBDesign Package).

Sample Size Parameters

Sample Size Parameters

Background on NMSC Incidence

  • Risk of subsequent NMSC is proportional to numbers of previously diagnosed skin cancers
  • In one study, the five year risk of developing another skin cancer was estimated at >60% for individuals with two previous skin cancers and at >90% for individuals with 4-5 previous skin cancers(Karagas 1992)
  • In ONTRAC, patients receiving placebo developed on average 2.4 NMSCs (0.7 CSCCs) in 12 months(Chen et al. 2015)
    • Inclusion: ≥2 NMSCs in the previous 5 years
    • Actual Baseline: 7.9 ± 8.0 NMSCs, 2.1 ± 3.5 CSCCs in Placebo
  • In ONTRANS, patients receiving placebo developed on average 2.7 NMSCs (1.9 CSCCs) in 12 months(Allen et al. 2023)
    • Inclusion: ≥2 NMSCs in the previous 5 years
    • Actual Baseline: 7.5 ± 7.6 NMSCs, 4.8 ± 5.6 CSCCs in Placebo

Eligibility Criteria

  • Inclusion
    • At least 10 NMSC, ≥5 CSCCs
    • ECOG 0, 1 or 2
  • Exclusion
    • Previous ICI
    • Active pharmacological immunosuppression
    • HIV with detectable viral loads (undetectable dz allowed)
    • Field treatment with the past 4 weeks
    • Nicotinamide use within the past 3 months

Calculating the sample sizes

[1] 254 259 258
[1] 254 259 258
[1] 213 216 216
[1] 217 221 220
             Outcome +    Outcome -      Total               Prevalence *
Exposed +          206           79        285     72.28 (66.69 to 77.40)
Exposed -          210           79        289     72.66 (67.14 to 77.72)
Total              416          158        574     72.47 (68.62 to 76.09)

Point estimates and 95% CIs:
-------------------------------------------------------------------
Prevalence ratio                               0.99 (0.90, 1.10)
Odds ratio                                     0.98 (0.68, 1.41)
Attrib prevalence in the exposed *             -0.38 (-7.69, 6.92)
Attrib fraction in the exposed (%)            -0.53 (-11.20, 9.11)
Attrib prevalence in the population *          -0.19 (-6.50, 6.11)
Attrib fraction in the population (%)         -0.26 (-5.40, 4.62)
-------------------------------------------------------------------
Uncorrected chi2 test that OR = 1: chi2(1) = 0.011 Pr>chi2 = 0.918
Fisher exact test that OR = 1: Pr>chi2 = 0.926
 Wald confidence limits
 CI: confidence interval
 * Outcomes per 100 population units 
       est    lower    upper
1 2.607595 2.263646 2.989036
2 2.658228 2.310837 3.043103


    Exact Poisson test

data:  5 time base: 3
number of events = 5, time base = 3, p-value = 0.2345
alternative hypothesis: true event rate is not equal to 1
95 percent confidence interval:
 0.5411621 3.8894440
sample estimates:
event rate 
  1.666667 

     Two-sample Negative Binomial rates Tests (Equal Sizes) 

              N = 175.1805
            mu1 = 4.9
            mu2 = 7
          theta = 0.8
       duration = 1
      sig.level = 0.05
          power = 0.8
    alternative = two.sided

NOTE: N is number in *each* group

Treatment Considerations

ICI Antibody Characteristics

  • For most ICIs (except ipilimumab), there is no clear relationship between dose and efficacy or safety
  • The dose-response and exposure-response curves showed an obvious plateau, possilby implying that increasing doses do not contribute to tumor control

ICI Antibody Characteristics

ICI Characteristics
Target Drug Half-life Dose Schedule
CTLA-4 Ipilimumab ~15 days 3 mg/kg q3 weeks
CTLA-4 Ipilimumab ~15 days 10 mg/kg q3 weeks
PD-1 Nivolumab ~25 days 240 mg q2 weeks
PD-1 Nivolumab ~25 days 480 mg q4 weeks
PD-1 Pembrolizumab ~25 days 200 mg q3 weeks
PD-1 Pembrolizumab ~25 days 400 mg q6 weeks
PD-1 Cemiplimab ~19 days 350 mg q3 weeks
PD-1 Retinfanlimab ~18 days 500 mg q4 weeks
PD-L1 Avelumab ~4 days 800 mg q2 weeks
PD-L1 Atezolizumab ~27 days 840 mg q2 weeks
PD-L1 Atezolizumab ~27 days 1200 mg q3 weeks
PD-L1 Atezolizumab ~27 days 1680 mg q4 weeks
PD-L1 Durvalumab ~18 days 10 mg/kg q2 weeks
PD-L1 Durvalumab ~18 days 1500 mg q4 weeks

ICI Dose Response

Dose-Efficacy
Target Drug Indication Dose ORR (%) PFS (% at 2 years) G3-G4 toxicity (%)
CTLA-4 Ipilimumab Melanoma 0.3 mg/kg Q3W 0 2.7
CTLA-4 Ipilimumab Melanoma 3 mg/kg Q3W 4.2-19 12.9 15-27.3
CTLA-4 Ipilimumab Melanoma 10 mg/kg Q3W 11.1-15 18.9 31-34
PD-1 Nivolumab Melanoma 0.1 mg/kg Q2W 29-35 40-41 0-5
PD-1 Nivolumab Melanoma 0.3 mg/kg Q2W 19-28 31-35 0-3
PD-1 Nivolumab Melanoma 1 mg/kg Q2W 30-31 45-51 6-12
PD-1 Nivolumab NSCLC 1 mg/kg Q2W 3-6 24-46 15.2
PD-1 Nivolumab NSCLC 3 mg/kg Q2W 24-32 40-41 13.5
PD-1 Nivolumab NSCLC 10 mg/kg Q2W 18-20 24-33 13.6
PD-1 Nivolumab RCC 0.3 mg/kg Q2W 20 30 18.2
PD-1 Nivolumab RCC 1 mg/kg Q2W 24-28 47-50 12
PD-1 Nivolumab RCC 2 mg/kg Q2W 22 30 17
PD-1 Nivolumab RCC 10 mg/kg Q2W 31 67 13
PD-1 Pembrolizumab Melanoma 2 mg/kg Q3W 21-32.9 45 8-15
PD-1 Pembrolizumab Melanoma 10 mg/kg Q2W 33.7-52 13.3-22.8
PD-1 Pembrolizumab Melanoma 10 mg/kg Q3W 26-35.9 37 3.6-15
PD-1 Pembrolizumab NSCLC 2 mg/kg Q3W 15-25 13
PD-1 Pembrolizumab NSCLC 10 mg/kg Q2W 19.3-21 9
PD-1 Pembrolizumab NSCLC 10 mg/kg Q3W 19.2-25 9

ICI Dose Response

Dose-Efficacy
Target Drug Indication Dose ORR (%) PFS (% at 2 years) G3-G4 toxicity (%)
CTLA-4 Ipilimumab Melanoma 0.3 mg/kg Q3W 0 2.7
CTLA-4 Ipilimumab Melanoma 3 mg/kg Q3W 4.2-19 12.9 15-27.3
CTLA-4 Ipilimumab Melanoma 10 mg/kg Q3W 11.1-15 18.9 31-34
PD-1 Nivolumab Melanoma 0.1 mg/kg Q2W 29-35 40-41 0-5
PD-1 Nivolumab Melanoma 0.3 mg/kg Q2W 19-28 31-35 0-3
PD-1 Nivolumab Melanoma 1 mg/kg Q2W 30-31 45-51 6-12
PD-1 Nivolumab NSCLC 1 mg/kg Q2W 3-6 24-46 15.2
PD-1 Nivolumab NSCLC 3 mg/kg Q2W 24-32 40-41 13.5
PD-1 Nivolumab NSCLC 10 mg/kg Q2W 18-20 24-33 13.6
PD-1 Nivolumab RCC 0.3 mg/kg Q2W 20 30 18.2
PD-1 Nivolumab RCC 1 mg/kg Q2W 24-28 47-50 12
PD-1 Nivolumab RCC 2 mg/kg Q2W 22 30 17
PD-1 Nivolumab RCC 10 mg/kg Q2W 31 67 13
PD-1 Pembrolizumab Melanoma 2 mg/kg Q3W 21-32.9 45 8-15
PD-1 Pembrolizumab Melanoma 10 mg/kg Q2W 33.7-52 13.3-22.8
PD-1 Pembrolizumab Melanoma 10 mg/kg Q3W 26-35.9 37 3.6-15
PD-1 Pembrolizumab NSCLC 2 mg/kg Q3W 15-25 13
PD-1 Pembrolizumab NSCLC 10 mg/kg Q2W 19.3-21 9
PD-1 Pembrolizumab NSCLC 10 mg/kg Q3W 19.2-25 9

Background

To model count data, we can use a poisson distribution, as count data are not continuous (they’re whole numbers), and thus a poisson regression. However, Poisson regression assumes that the variance of y equals the mean of y. A dispersion parameter of 1 means that that mean = variance. However, in many real-world data sets, the variance is greater than the mean, and this is called “overdispersion”. A hint of this is if the residual deviance in your model is equal to the degrees of freedom. Dispersion Parameter == Sum of Squared Pearson Residuals / df. Overdispersion can result in underestimation of the standard error (more likely to have a false positive). Causes of overdispersion include predictor variables that are not included in the model; clustering or heterogeneity in the sampled population.

Negative binomial model is a poisson model but allowing parameter k. Variance = mean + mean2/k. Only works for overdispersion, not underdispersion. (can also choose a quasi-poisson)

Alterative Approach

Inclusion Criteria

Open ICARUS

Inclusion Criteria

References

Allen, Nicholas C., Andrew J. Martin, Victoria A. Snaidr, Renee Eggins, Alvin H. Chong, Pablo Fernandéz-Peñas, Douglas Gin, et al. 2023. “Nicotinamide for Skin-Cancer Chemoprevention in Transplant Recipients.” New England Journal of Medicine 388 (9): 804–12. https://doi.org/10.1056/nejmoa2203086.
Chen, Andrew C., Andrew J. Martin, Bonita Choy, Pablo Fernández-Peñas, Robyn A. Dalziell, Catriona A. McKenzie, Richard A. Scolyer, et al. 2015. “A Phase 3 Randomized Trial of Nicotinamide for Skin-Cancer Chemoprevention.” New England Journal of Medicine 373 (17): 1618–26. https://doi.org/10.1056/nejmoa1506197.
Karagas, Margaret R. 1992. “Risk of Subsequent Basal Cell Carcinoma and Squamous Cell Carcinoma of the Skin Among Patients With Prior Skin Cancer.” JAMA: The Journal of the American Medical Association 267 (24): 3305. https://doi.org/10.1001/jama.1992.03480240067036.