Table 4 Details regarding cost-effectiveness studies of melanoma

Barzey et al., USA, 2013 [37]

Ipi vs. BSC for recurrent/metastatic disease

Markov; accounted for toxicity and administration costs

$168,602 ipi, $21,886 BSC

1.76 ipi, 0.62 BSC

Relative to BSC, ipi $128,656/QALY

$146,000/QALY

Ipi CE using given WTP threshold, not so at more accepted cutoffs

- Overall modeling horizon of complete lifetime, causing errors in survival extrapolation and thus costs

- Lack of accountability for hospice/palliative care and death costs

- BSC arm (without chemotherapy) provides little meaningful clinical comparison

Curl et al., USA, 2014 [38]

Dac vs. vem vs. vem + ipi for unresected/metastatic BRAF mutant disease

Deterministic expected-value model; accounted for toxicity, administration, and follow-up costs

$8391 dac, $156,831 vem, $254,695 vem + ipi

0.30 dac, 0.72 vem, 1.34 vem + ipi

Relative to dac, vem $353,993/QALY, vem + ipi $158,139/QALY

No specific amount used

Vem or vem + ipi not CE in this setting

- Overall modeling horizon of complete lifetime, causing errors in survival extrapolation and thus costs

- Assumed effect of vem + ipi is seamlessly additive

- Lack of accountability for hospice/palliative care and death costs; unclear methodology for toxicity costs

Bohensky et al., Australia, 2016 [39]

Nivo vs. ipi for unresected/metastatic BRAF WT disease

Markov; accounted for toxicity, administration, and end-of-life costs

$178,612 nivo, $138,987 ipi

2.5 nivo, 1.2 ipi

Relative to ipi, nivo $30,475/QALY

$35,000/QALY

Nivo is more CE than ipi in this setting

- Overall modeling horizon of 10 years, causing errors in survival extrapolation and thus costs

- Used data from second-line ipi and extrapolated to first-line ipi

- Assumed patients weigh same as mean Australian body weight in trial (dosed accordingly); duration of therapy assumed to be same as the mean amount on trial

Oh et al., USA, 2017 [40]

Nivo vs. ipi vs. nivo+ipi for unresected/metastatic disease

Markov; accounted for toxicity, administration, follow-up, and end-of-life costs

$169,320 nivo, $213,763 ipi, $228,352 both

4.24 nivo, 3.68 ipi, 4.37 both

Relative to nivo, ipi was dominated; relative to ipi, both $21,143/QALY; relative to nivo, both $454,092/QALY

$100,000/QALY

Nivo (single-agent) is most CE in this setting; PD-L1 status changes cost-effectiveness negligibly

- Overall modeling horizon of 14.5 years, causing errors in survival extrapolation and thus costs

- Owing to no overall survival data, survival figures were dependent on progression-free survival values only

- Did not account for 2nd or 3rd line therapies

Wang et al., USA, 2017 [41]

Pembro vs. ipi for unresected/metastatic disease

PS; accounted for toxicity, administration, follow-up, and end-of-life costs

$303,505 pembro, $239,826 ipi

3.45 pembro, 2.67 ipi

Relative to ipi, pembro $81,091/QALY

$100,000/QALY

Pembro is more CE than ipi in this setting

- Overall modeling horizon of 20 years, causing errors in survival extrapolation and thus costs

- Assumed no systemic therapy of any kind following progression

- Pembro planned for maximum of 24 months in model (instead of until progression)

Miguel et al., Portugal, 2017 [42]

Pembro vs. ipi for unresected/metastatic disease

PS; accounted for toxicity, administration, and end-of-life costs

€156,268 ($191,924) pembro, €110,034 ($135,140) ipi

3.31 pembro, 2.33 ipi

Relative to ipi, pembro €47,221 ($57,988)/QALY

€50,000 ($61,407)/QALY

Pembro is more CE than ipi in this setting

- Overall modeling horizon of 40 years, when exceedingly low numbers of patients still alive, causing errors in survival extrapolation and thus costs

- Only grade ≥ 3 toxicities accounted for, as a one-time cost

- Pembro planned for maximum of 24 months in model (instead of until progression)

Kohn et al., USA, 2017 [43]

Dac vs. nivo vs. ipi vs. nivo+ipi vs. pembro (q2w) vs. pembro (q3w) for unresected/metastatic disease

Markov with built-in transition to 2nd and 3rd line therapies; accounted for toxicity, administration, and end-of-life costs

$146,775 dac, $172,219 nivo, $152,403 ipi, $206,435 nivo+ipi, $164,871 q2w pembro, $127,626 q3w pembro

0.26 dac, 0.54 nivo, 0.34 ipi, 0.56 nivo+ipi, 0.43 q2w pembro, 0.38 q3w pembro

Relative to q3w pembro, dac, ipi, and q2w pembro were dominated; nivo $66,800/QALY; nivo+ipi $319,723/QALY

$100,000/QALY

Nivo or q3w pembro (followed by 2nd line ipi) is most CE in this setting

- Overall modeling horizon of complete lifetime, causing errors in survival extrapolation and thus costs

- No prospective data for several arms (e.g. pembro followed by 2nd line ipi)

- Did not use immunotherapy dosing by body weight; although ongoing trials may not utilize weight-based dosing, previous trials (i.e., major sources of extracted data) have done so

Meng et al., England, 2018 [44]

Dac vs. ipi vs. nivo for unresected/metastatic BRAF WT; ipi vs. dab vs. vem vs. nivo for BRAF mutant disease

Markov; accounted for toxicity, administration, and end-of-life costs

BRAF WT: dac £25,228 ($35,542), ipi £57,158 ($80,532), nivo £97,898 ($137,931); BRAF mutant: ipi £56,621 ($79,775), dab £71,511 ($100,754), vem £74,001 ($104,262), nivo £88,228 ($124,307)

1.23 dac, 2.54 (avg) ipi, 1.69 dab, 1.70 vem, 4.29 (avg) nivo

BRAF WT: relative to dac, ipi £22,589 ($31,825)/QALY, nivo £24,483 ($34,493)/QALY; BRAF mutant: relative to ipi, dab and vem dominated; nivo £17,362 ($24,460)/QALY

£50,000 ($70,462)/QALY

Nivo is most CE in these settings

- Overall modeling horizon of complete lifetime, causing errors in survival extrapolation and thus costs

- Model sensitive to treatment duration, but nonuniform comparison of continuing nivo for 2 years versus dac and vem until progression

- Unclear description of cost summary with 2nd line of therapy