1. Systematic Literature Review

Appendices

1. Systematic literature review

A systematic review of the published literature was conducted to identify randomised controlled trial (RCT) evidence on the efficacy and safety of interventions for the management of AK lesions.

Initially, there were no restrictions on comparator treatments. This systematic review used broad search terms to identify all existing comparators and outcomes. Studies analysing comparators not of relevance to Scottish clinical practice and thus the economic model were subsequently excluded from further analysis.

The Cochrane Library, OVID Medline in process, OVID Medline and OVID EMBASE were searched for relevant data. Using Boolean operators, the searches combined terms (including MeSH headings as appropriate) for AK, pharmacological intervention(s) of interest, and clinical trial design. Searches were originally performed on 9th May 2011 and the review updated on 14th May 2012.

The following databases were searched using the search strategies detailed subsequently:

 The Cochrane Library, incorporating;

o Cochrane Database of Systematic Reviews (Cochrane Reviews)

o Database of Abstracts of Reviews of Effects (DARE)

o Cochrane Central Register of Controlled Trials (CENTRAL)

o Health Technology Assessment Database (HTA)

o NHS Economic Evaluation Database (NHS EED)

 OVID Medline in process

 OVID Medline, 1950 to present day

 OVID EMBASE, 1980 to present day

The search strategy combined AK Mesh term (actinic keratosis) with free text treatment terms and was adapted to search each database.

Inclusion and exclusion selection criteria are shown in Table 11. The systematic review schematic is presented in Figure 6. Articles identified were initially assessed based on title and abstract. Papers not meeting the inclusion criteria were excluded and allocated a

“reason code” to document the rationale for exclusion. Papers included after this stage were then assessed based on the full text; further papers were excluded, yielding the final data set for inclusion. The final included data set from the RCT search consisted of clinical studies for ingenol mebutate gel and those for comparator treatments.

Following assessment and exclusion based on title, abstract and full text, 104 articles were identified covering all AK treatments. A further exclusion step was performed in order to capture articles reporting on treatments of relevance to Scotland for inclusion into a mixed treatment comparison

(MTC). In total, 94 articles were excluded. Studies were excluded for the following reasons: treatment not relevant (n=81), patients not relevant (n=3), outcomes not relevant (n=5). In addition, two abstracts were replaced by a full publication identified in the 2012 search and three studies were found to have ambiguous data reporting making them unsuitable for inclusion in the MTC.

A senior analyst conducted the searches. Potentially relevant studies (based on abstract/title) were ordered and examined in full by the analyst and verified by a project manager. Disagreements as to eligibility were referred to a third party (project director).

2. Mixed treatment comparison

An MTC was conducted to estimate the relative effectiveness of ingenol mebutate gel and comparator treatments. The MTC methodology and code was as recommended by the NICE Decision

Support Unit [1-3]. The model was coded in WinBUGS software version 1.4.3 [4]; the WinBUGS code is as follows: model{ for(i in 1:NS){

Temp1[i]<-Study[i] # unused data

w[i,1] <-0

delta[i,t[i,1]]<-0 mu[i] ~ dnorm(0,.0001) # vague priors for trial baselines

for (k in 1:na[i]) {

r[i,k] ~ dbin(p[i,t[i,k]],n[i,k]) # binomial likelihood

logit(p[i,t[i,k]])<-mu[i] + delta[i,t[i,k]] # model

# Estimation of deviances rhat[i,k] <- p[i,t[i,k]] * n[i,k] dev[i,k] <- 2 * (r[i,k] * (log(r[i,k])-log(rhat[i,k])) + (n[i,k]-r[i,k]) *(log(n[i,k]-r[i,k]) - log(n[i,k]-rhat[i,k])))

} deva[i]<- sum(dev[i,1:na[i]])

for (k in 2:na[i]) { delta[i,t[i,k]] ~ dnorm(md[i,t[i,k]],taud[i,t[i,k]]) # trial-specific LOR dist md[i,t[i,k]] <- d[t[i,k]] - d[t[i,1]] + sw[i,k]

# + beta * (equals(t[i,1],1)) * (Covariate[i] -mean(Covariate[]))

# + beta2 * (equals(t[i,1],1)) * (Covariate2[i] -mean(Covariate2[])) taud[i,t[i,k]] <- tau *2*(k-1)/k #precision of LOR distributions w[i,k] <- (delta[i,t[i,k]] - d[t[i,k]] + d[t[i,1]]) #adjustment, multi-arm RCTs sw[i,k] <-sum(w[i,1:k-1])/(k-1) } # cumulative adjustment for multi-arm trials

} sumdev<- sum(deva[])

#beta~dnorm(0,0.0001)

#beta2~dnorm(0,0.0001) d[1]<-0 for (k in 2:NT){d[k] ~ dnorm(0,.0001) } # vague priors for basic parameters sd~dunif(0,2) # vague prior for random effects standard deviation tau<-1/pow(sd,2)

# Absolute log odds(success) on Treatment A, based on a separate model/data source. mA ~ dnorm(meaMTC,precmA) for (k in 1:NT) { logit(T[k])<- mA +d[k] }

# ranking for (k in 1:NT) { logit(Trk[k])<- mA +d[k] rk[k]<-NT+1 - rank(Trk[],k) best[k]<-equals(rk[k],1)}

# pairwise ORs for (c in 1:(NT-1))

{ for (k in (c+1):NT)

{ lor[c,k] <- d[k] - d[c]

log(or[c,k]) <- lor[c,k]

The MTC results were based on two chains of 20,000 samples (after a minimum burn in of 100,000 samples). The random-effects model appeared to fit the data better than the fixed-effect model: the average residual deviance was 1.446 and 1.040, and the DIC was 151.291 and 145.143 for the fixed- effect and random-effects NMA respectively.

The between study deviance estimated by the random-effects model was 0.89 which indicates that there is substantial between study heterogeneity. As a consequence of study heterogeneity the distribution of odds ratios observed from the trials will have a range (between 95% upper and lower limits) of 32.61 around the true mean treatment effect.

The primary clinical outcome analysed in the MTC was complete clearance, defined as the number of patients with no clinically visible AK lesions in the treatment area. Some studies reported this as target lesion number score (TLNS), defined as the proportion of lesions completely healed (zero lesions) after treatment. The analysis was conducted on an intent-to-treat basis: number of patients with complete clearance at follow-up as a proportion of number randomised to the treatment. For participants lost to follow-up, it was assumed that patients whose clearance status was unknown at follow-up did not achieve complete clearance. Figure 7 shows the network diagram for studies reporting the relevant complete clearance data and shows the results from these studies that were used to inform the MTC.

Eleven independent studies were included in the MTC. Placebo controlled studies or studies reporting data for imiquimod 5% 3x/week for 4 weeks were included in the network. Although not considered a relevant comparator for the model, inclusion of imiquimod and placebo-controlled studies was necessary in order to connect all seven of the treatments of relevance to Scottish clinical practice.

One publication, reporting data for ingenol mebutate gel [5], provided an analysis of the four pivotal trials for ingenol mebutate gel. The two RCTs that investigated the efficacy and safety of ingenol mebutate gel (150mcg/g) on the face/scalp were combined into a single analysis of efficacy and safety for this strength. Similarly, the two RCTs that investigated the efficacy and safety of ingenol mebutate gel (500mcg/g) on the trunk or extremities were combined into a single analysis of efficacy and safety. In the systematic review and MTC, this publication is considered as two studies; to reflect the combined data for ingenol mebutate 150mcg/g gel and ingenol mebutate 500mcg/g gel, whilst keeping these strengths (and therefore the locations treated) separate.

Although imiquimod was not considered a relevant comparator for the analysis, it was included in the MTC as it provided the relevant links to cryotherapy and 5-FU.

Patients with AK lesions on the following locations were included in each of the studies:

• Akarsu et al, 2011 – face and arms

• Alomar et al, 2007 – face or scalp

• Anderson et al, 2009 arm, shoulder, chest, back, or scalp • Gebauer et al, 2003 - head, neck, hands and arms

• Jorizzo et al, 2007 -face and scalp

• Krawtchenko et al, 2007 – head, neck and décolleté

• Lebwohl et al, 2012a – face and scalp (ingenol mebutate 150 mcg/g)

• Lebwohl et al, 2012b – trunk and limbs (ingenol mebutate 500 mg/g)

• Rivers et al, 2002 - face scalp and hands

• Stockfleth et al, 2011 – face and scalp

• Wolf et al, 2001 – face, scalp, arms, hands

The LOR were exponentiated and then applied to the baseline odds of CC with vehicle. This was achieved by multiplying the odds of CC by the OR with the associated treatment. The odds were then converted to probabilities of CC for use as data inputs in the model. If the odds in favour of an event are known, the probability is calculated as follows:

3. Percentage of patients experiencing AEs

Table 13 presents details of the number of patients that experienced AEs in the studies included in the MTC.

4. Supporting model results

In univariate sensitivity analyses of ingenol mebutate 150 mcg/g and 500 mcg/g vs diclofenac (8 weeks), the parameters that had the largest impact on the ICER were the utility value for complete clearance, the proportion of patients referred to secondary care for each therapy and the utility value for AK (Figure 8). In univariate sensitivity analyses of ingenol mebutate 150 mcg/g and 500 mcg/g vs 5-FU, the parameters that had the largest impact on the ICER were the utility values for complete clearance and AK; and the proportion of patients referred to secondary care for each therapy (Figure 9).

In univariate sensitivity analyses of ingenol mebutate 150 mcg/g and 500 mcg/g gel vs 5-FU/salicylic acid, the parameters that had the largest impact on the ICER were the utility values for complete clearance, the proportion of patients referred to secondary care for each treatment and the utility value for AK (Figure 10).

In univariate sensitivity analyses of ingenol mebutate 150 mcg/g and 500 mcg/g gel vs cryotherapy, the parameters that had the largest impact on the ICER were the utility values for AK and complete clearance and the proportion of patients referred to secondary care to receive ingenol mebutate gel

(Figure 11). Appendix figure legends

Fig.6 Systematic review flow diagram

Fig.7 Network diagram for complete clearance outcome

Abbreviations: 5-FU, fluorouracil; SA, salicylic acid.

*plus second course of treatment as needed; NB: blue shading –included in order to connect all treatments of interest

Fig.8 Tornado diagrams of univariate sensitivity analyses for ingenol mebutate gel 150 mcg/g for 3 days (a) and 500 mcg/g for 3 days (b) vs diclofenac (3%) for 12 weeks

Abbreviations: AK, actinic keratosis; ICER, incremental cost-effectiveness ratio. * Truncated for illustrative purposes.

Fig.9 Tornado diagrams of univariate sensitivity analyses for ingenol mebutate gel 150 mcg/g for 3 days (a) and 500 mcg/g for 3 days (b) vs 5-FU (5%) 3 x daily for 4 weeks

Fig.10 Tornado diagrams of univariate sensitivity analyses for ingenol mebutate gel 150 mcg/g for 3 days (a) and 500 mcg/g for 3 days (b) vs 5-FU (0.5%)/salicylic acid (10%) 1 x daily for 12 weeks

Fig. 11 Tornado diagrams of univariate sensitivity analyses for ingenol mebutate gel 150 mcg/g for 3 days (a) and 500 mcg/g for 3 days (b) vs cryotherapy

Abbreviations: AK, actinic keratosis; ICER, incremental cost-effectiveness ratio. * Truncated for illustrative purposes. Appendix references

1. Dias, S., Welton, N., Sutton, A., Ades, A.E.: NICE DSU technical support document 2: A generalised linear modelling framework for pairwise and network meta-analyisis of randomised controlled trials. Available at http://www.nicedsu.org.uk/TSD2%20General %20meta%20analysis.final.08.05.12.pdf. (2011). 2. Dias, S., Welton, N.J., Sutton, A.J., Ades, A.E.: NICE DSU technical support document 5: Evidence synthesis in the baseline natural history model. Available at http://www.nicedsu.org.uk/TSD5%20Baseline.final%20report.08.05.12.pdf. (2011). 3. Welton, N.J., Sutton, A.J., Cooper, N.J., Abrams, K.R., Ades, A.E.: Evidence Synthesis for Decision Making in Healthcare, First ed. Statistics in Practice. John Wiley & Sons, Ltd, Chichester, UK (2012) 4. Swanson, N., Melgaard, A., Larsson, T.: Multicenter, randomized, parallel-group, double- blind, vehicle-controlled phase 3 study to evaluate the efficacy and safety of PEP005 (ingenol mebutate) gel, 0.05% in patients with actinic keratoses on non-head locations (study PEP005-014). Poster presented at: 22nd World Congress of Dermatology, Seoul, Korea, Poster P2182 (2011). 5. Lebwohl, M., Swanson, N., Anderson, L.L., Melgaard, A., Xu, Z., Berman, B.: Ingenol mebutate gel for actinic keratosis. N Engl J Med 366(11), 1010-1019 (2012). 6. Akarsu, S., Aktan, S., Atahan, A., Koc, P., Ozkan, S.: Comparison of topical 3% diclofenac sodium gel and 5% imiquimod cream for the treatment of actinic keratoses. Clin Exp Dermatol 36(5), 479-484 (2011). 7. Alomar, A., Bichel, J., McRae, S.: Vehicle-controlled, randomized, double-blind study to assess safety and efficacy of imiquimod 5% cream applied once daily 3 days per week in one or two courses of treatment of actinic keratoses on the head. Br J Dermatol 157(1), 133-141 (2007). 8. Anderson, L., Schmieder, G.J., Werschler, W.P., Tschen, E.H., Ling, M.R., Stough, D.B., Katsamas, J.: Randomized, double-blind, double-dummy, vehicle-controlled study of ingenol mebutate gel 0.025% and 0.05% for actinic keratosis. J Am Acad Dermatol 60(6), 934-943 (2009). 9. Gebauer, K., Brown, P., Varigos, G.: Topical diclofenac in hyaluronan gel for the treatment of solar keratoses. Australas J Dermatol 44(1), 40-43 (2003). 10. Jorizzo, J., Dinehart, S., Matheson, R., Moore, J.K., Ling, M., Fox, T.L., McRae, S., Fielder, S., Lee, J.H.: Vehicle-controlled, double-blind, randomized study of imiquimod 5% cream applied 3 days per week in one or two courses of treatment for actinic keratoses on the head. J Am Acad Dermatol 57(2), 265-268 (2007). 11. Krawtchenko, N., Roewert-Huber, J., Ulrich, M., Mann, I., Sterry, W., Stockfleth, E.: A randomised study of topical 5% imiquimod vs. topical 5-fluorouracil vs. cryosurgery in immunocompetent patients with actinic keratoses: a comparison of clinical and histological outcomes including 1-year follow-up. Br J Dermatol 157 Suppl 2, 34-40 (2007). 12. Rivers, J.K., Arlette, J., Shear, N., Guenther, L., Carey, W., Poulin, Y.: Topical treatment of actinic keratoses with 3.0% diclofenac in 2.5% hyaluronan gel. Br J Dermatol 146(1), 94-100 (2002). 13. Stockfleth, E., Kerl, H., Zwingers, T., Willers, C.: Low-dose 5-fluorouracil in combination with salicylic acid as a new lesion-directed option to treat topically actinic keratoses: histological and clinical study results. Br J Dermatol 165(5), 1101-1108 (2011). 14. Wolf, J.E., Jr., Taylor, J.R., Tschen, E., Kang, S.: Topical 3.0% diclofenac in 2.5% hyaluronan gel in the treatment of actinic keratoses. Int J Dermatol 40(11), 709-713 (2001).