Molecular Lessons Learnt from

Clinical Trials in the Last Decade

Andrew X. Zhu, MD, PhD

Harvard Medical School
Massachusetts General Hospital Cancer Center
Jiahui International Cancer Center
 Disclosure

• Advisory/consulting: Eisai, Bristol-Myers Squibb, Merck, Novartis,

Astra-Zeneca, Bayer, Lilly, Exelixis, Roche-Genentech

• Research funding: Merck, Novartis, Bristol-Myers Squibb, Bayer,

Lilly
 Discussion Points
• Progress has been made in advanced HCC through well conducted clinical trials
• Lessons learned from clinical trials
• Challenges in trial design with targeted agents and immunotherapy
– Biomarker enrichment trials

– Response assessment

– Endpoints selection

• Mechanism of drug resistance

• Emerging targets and strategies in advanced HCC

3
 Shifting Therapeutic Landscape of NSCLC
Pembro +
Nivo Pembro Atezo chemo Durva
Mar 2015 Oct 2015 Oct 2016 May 2017 Feb 2018
Crizotinib
Mar 2016

Crizotinib Ceritinib Alectinib Brigatinib Lorlatinib

Aug 2011 April 2014 Dec 2015 April 2017 July 2018

2011 2012 2013 2014 2015 2016 2017 2018

Erlotinib Afatinib Gefitinib Osimertinib

May 2013 July 2013 July 2015 Nov 2015

Ramucirumab Necitumumab
Dec 2014 Nov 2015

4
 Improving Long-Term Survival in Lung Cancer
ALK+ NSCLC (PROFILE 1014)
100
+ Censored
Median follow-up ~46 months in both arms

80
HR 0.760 (95%CI: 0.548, 1.053);
a
P=0.0978
Overall Survival (%)

60

40

Crizotinib Chemotherapy
(N=172) (N=171) 4 year OS rate
20
Deaths, n (%) 71 (41.3) 81 (47.4) Crizo: 56.6%
Median OS (95% CI), months NR (45.8, NR) 47.5 (32.2, NR) Chemo: 49.1%
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

Months
No. at risk
Crizotinib 172 157 144 128 111 98 89 79 65 51 36 20 8 1 0
Chemotherapy 171 150 131 118 100 89 82 73 63 46 31 21 11 1 0
5
Mok et al., ESMO 2017
 The Evolving Treatment Landscape for Advanced HCC
Nivolumab
Approved for 2L HCC September 2017 (accelerated)
Phase 3 Checkmate 459 fails to meet
primary endpoint in 1L uHCC (June 2019)

Pembrolizumab
Approved for 2L HCC November 2018 (accelerated)
Phase 3 Keynote-240 fails to meet
co-primary endpoints in 2L HCC (February 2019) 7

2005 2010 2015 2020

Ramucirumab
Approved for 2L HCC
Regorafenib2 (AFP >400 ng/mL) May 2019
Sorafenib Approved for 2L HCC April 2017
Approved for 1L HCC November 2007

Lenvatinib
Approved for 1L HCC August 2018
Keynote 240: Median OS 13.9 months in second line HCC
Checkmate 459: Median OS in first line: ?
Cabozantinib
6 Approved for 2L HCC January 2019
 Sorafenib in advanced HCC
SHARP vs Asia-Pacific Study: Overall Survival
SHARP Trial Asia-Pacific Trial

P=0.014

10.7 vs 7.9 mo 6.5 vs 4.2 mo

HR: 0.69 (0.55-0.87) HR: 0.68 (0.50-0.93)
7 1. Llovet JM, et al. N Engl J Med 2008 359:378-90
2. Cheng AL, et al. Lancet Oncology 2009
 Failed Phase III Trials in Advanced HCC
Molecular targets Median survival,
Trial Drugs Design N HR P Value
(signaling pathways) mo
First-line treatment
Sorafenib 299 10.7
SHARP[a] BRAF, VEGFR, PDGFR Superiority 0.69 .001
Placebo 303 7.9
Sorafenib 150 6.5
Asia-Pacific[b] BRAF, VEGFR, PDGFR Superiority 0.68 .01
Placebo 76 4.2
Sunitinib 530 7.9
Sunitinib[c] c-Kit, VEGFR, PDGFR Superiority 1.3 .001
Sorafenib 544 10.2
Brivanib FGFR, VEGFR 577 9.5
BRISK-FL[d] Noninferiority 1.06 .31
Sorafenib BRAF, VEGFR, PDGFR 578 9.9
Linifanib VEGFR, PDGFR 9.1
LIGHT[e] Noninferiority 1035 1.04 .52
Sorafenib BRAF, VEGFR, PDGFR 9.8
Sorafenib+erlotinib EGFR, BRAF, VEGFR, PDGFR 362 9.5
SEARCH[f] Superiority 0.93 .2
Sorafenib BRAF, VEGFR, PDGFR 358 8.5
CALGB Sorafenib+doxorubicin -- 180 8.9
Superiority 1.06 .24
80802[g] Sorafenib BRAF, VEGFR, PDGFR 176 10.5
FOLFOX 4 -- 184 6.4
EACH[h] Superiority .80 .07
Doxorubicin -- 187 5.0
Second-line treatment
Brivanib 263 9.4
BRISK-PS[i] FGFR, VEGFR Superiority 0.89 .33
Placebo 132 8.2
Everolimus 362 7.6
EVOLVE-1[j] mTOR Superiority 1.05 .68
Placebo 184 7.3
Ramucirumab 277 9.2
REACH[k] VEGFR2 Superiority 0.86 .13
Placebo 276 7.6
ADI-PEG 20[l] 424 7.8
-- -- Superiority --- .884
Placebo 211 7.4

8 a. Llovet JM, et al. N Engl J Med. 2008;359:378-390; b. Cheng A-L, et al. Lancet Oncology. 2009;10:25-34; c. Cheng A-L, et al. J Clin Oncol. 2013; 31:4067-4075; d. Johnson PJ, et al. J Clin Oncol. 2013;31:3517-3524; e.
Cainap C, et al. J Clin Oncol. 2015;33:172-179; f. Zhu AX, et al. J Clin Oncol. 2015;33:559-566; g. Abou-Alfa GK, et al. ASCO 2016. Abstract 192; h. Qin S, et al. Oncologist. 2014;19:1169-1178; i. Llovet JM, et al. J Clin
Oncol. 2013;31:3509-3516; j. Zhu AX, et al. JAMA. 2014;312:57-67; k. Zhu AX, et al. Ann Oncol. 2016;27(suppl 6):vi207-vi242; l. Abou-Alfa GK, et al. ASCO 2016. Abstract 4017.
 Clinical Efficacy of Approved Agents in Advanced HCC
Overall Median
Median OS Median OS
Drugs (Study) response rate PFS/TTP
(months) (Hazard ratio)
(ORR, %) (months)

Sorafenib
2.3 5.5 10.7 0.69
(SHARP)
Lenvatinib
24.1 7.4 13.6 0.92
(REFLECT)
Regorafenib
10.6 3.1 10.6 0.63
(RESORCE)
Cabozantinib
3.8 5.5 10.2 0.76
(CELESTIAL)
Ramucirumab
4.6 2.8 8.5 0.71
(REACH-2)

9 a. Llovet JM, et al. N Engl J Med. 2008;359:378-390; b. Kudo M, et al. Lancet 2018. c. Bruix J, et al. Lancet.
2017;389:56-66; d. Abou-Alfa G, et al. N Engl J Med. 2008; e. Zhu AX, et al. Lancet Oncology 2019
 Checkpoint Inhibitors in Second-Line
Nivolumab Pembrolizumab
Sample size 154 sorafenib-treated patients 104 sorafenib-treated patients

Patient 2L or 3L 2L
features Sorafenib-intolerants allowed Sorafenib-intolerants allowed
Effective therapy for HBV+ve patients Effective therapy for HBV+ve patients
No involvement of portal vein trunk
Response rate 14% 17%
regardless of etiology or AFP levels regardless of etiology or AFP levels
Duration of 16.6 months in HCV patients, not reached in Not reached; ≥9 months in 77%
response other etiologies

mOS 15.1 months (95% CI 13.2–18.8) 12.9 months (95% CI 9.7-15.5)

El-Khoueiry AB et al. Lancet. 2017. pii: S0140-6736(17)31046-2

Zhu AX et al, Lancet Oncology 2018
 CheckMate-459: Nivolumab vs Sorafenib
• CHECKMATE-459: Phase 3 trial of nivolumab vs sorafenib in first-line advanced HCC
– Start Date: November 2015
– Primary endpoint: OS
– Other endpoints: ORR, PFS, TTP, RR biomarkers

Key Eligibility
Criteria
• Advanced HCC not
eligible for or Nivolumab
progressive after
surgical and/or R
locoregional
therapies Sorafenib
• Child-Pugh A
• (n = 726)

Press release (6/24/19): “did not achieve statistical significance OS primary endpoint
ClinicalTrials.gov. NCT02576509.
11 per the pre-specified analysis (HR 0.85, 95%CI 0.72-1.02, p=0752)”
 KEYNOTE-240 Study Design
Pembrolizumab
Key eligibility criteria 200 mg Q3W +BSC
− ≥18 y N=278
− Pathologically/radiographically confirmed HCC
− Progression on/intolerance to sorafenib
− Child Pugh class A Randomized 2:1*
− BCLC stage B/C
− ECOG PS 0-1
− Measurable disease per RECIST v1.1 Saline-placebo
− Main portal vein invasion was excluded Q3W +BSC
N=135

Stratification Factors
− Geographic region (Asia w/o Japan vs non-Asia
w/Japan)
− Macrovascular invasion (Y vs N)
− AFP level (≥200 vs <200 ng/mL)
12
 Overall Survival Objective Response Rate
100 Events HR (95% CI) P
Pembrolizumab 183 0.781 (0.611-0.998) 0.0238 13.8 (7.7-19.5)
90
Placebo 101 p=0.00007a
80
30
70 • Efficacy boundaries 18.3
Overall survival (%)

(14.0-23.4)
– p=0.0174 for OS

ORR, % (95% CI)

60
20
50
4.4
40
(1.6-9.4)
10
30 Median (95% CI)
13.9 mo (11.6-16.0)
20
10.6 mo (8.3-13.5) 0
10 Pembro Placebo
0
0 4 8 12 16 20 24 28 32
No. at risk Time (months) Median duration of response mon (range):
278 237 190 152 110 57 16 1 0 • pembrolizumab, 13.8 (1.5+ −23.6+)
135 113 84 65 42 23 8 1 0 • placebo, not reached (2.8−20.4+)
13
Data Cutoff: Jan 2, 2019. Finn R et al, ASCO 2019
 What Lessons We Have Learned from the Clinical Trials
• Assessing antitumor activity in early clinical trials is critical but challenging
• Patient selection is important
• Validated therapeutic targets: VEGFR and immune checkpoints
• Oncogene addition theory has not been successfully proven in HCC
• Molecular predictors of responses and clinical benefits are lacking
• Tumor heterogeneity
• Clinical benefits to TKIs may vary depending on the underlying etiology
• Pattern of progression might be important in predicting clinical outcome
• Side effects might be associated with clinical benefits
• How to choose and sequence different drugs?
14
 Assessing Antitumor Activity in Early Clinical
Trials Is Critical but Challenging
ClinicalTrials.gov:
• 1793 HCC trials registered (1353 trials with drugs)
– 424 phase I trials listed
– 707 phase II trials listed
– 180 randomized phase II trials listed
– 261 phase III trials listed
– Only 5 positive phase III trials

15
 Selection of Patients in Advanced HCC Trials

• ECOG PS: 0 or 1
• Macrovascular invasion: extent of vascular invasion VP4
• Extent of tumor involvement in the liver: 50%
• Underlying liver function: Child Pugh A (B7)

16
 Therapeutic Targeting of the Hallmarks of Cancer

17 Hanahan and Weinberg, Cell, 2011

Genome Sequencing in HCC

Guichard et al, Nat Gen 2012

Villanueva et al, Nat Rev Clin Oncol 2014
The Landscape of Altered Genes and Pathways in HCC

19
Schulze et al, Nature Genetics, 2015
 Tumor Microenvironment

limit

20 Courtesy of Dr. Lance L. Munn

 Multikinase Inhibitors and Antiangiogenic Agents
Positive Phase 3 Trials in HCC

Sorafenib[a] Oral TKI targeting RAF-1 and BRAF, VEGFR1–3, PDGFRβ

Lenvatinib[b] Oral TKI targeting VEGFR1–3, FGFR1–4, RET, KIT, PDGFRα

Regorafenib[c] Oral TKI targeting VEGFR1–3, TIE2, PDGFRβ, KIT, RET, RAF

Cabozantinib[d] Oral TKI targeting MET, AXL, and VEGFR

Ramucirumab[e] Human IgG1 monoclonal antibody against VEGFR2

a. Llovet JM, et al. N Engl J Med. 2008;359:378-390; b. Kudo M, et al. Lancet 2018. c. Bruix J, et al. Lancet. 2017;389:56-66; d. Abou-
21 Alfa G, et al. N Engl J Med. 2008; e. Zhu AX, et al. Lancet Oncology 2019
 Biomarkers for Sorafenib
• Tissue Biomarkers
– Nuclear pERK overexpression associated with prolonged TTP in phase 2 [1]
– Tissue pERK staining was not associated with outcomes in phase 3 [2]

• Circulating Biomarkers
– High s-c-Kit and low HGF at baseline showed a trend towards improved OS [3]

• Genomic Biomarkers
• FGF3/FGF4 and VEGF A amplification predicted response in small number
of HCC patients [4, 5]

1. Abou-Alfa GK, et al. J Clin Oncol. 2006;24:4293-4300. 2. Llovet JM, et al. N Engl J Med. 2008;359:378-390. 3. Llovet JM, et al. Clin Cancer Res. 2012;18:2290-
22 2300. 4. Arao T, et al. Hepatology, 2013; 57(4):1407-15; Horwitz E at al, Cancer Discoery, 2014.
FGF3/FGF4 Amplification Predicting Sorafenib Sensitivity

23
Arao et al, Hepatology 2013
 Proteins Identified as Potentially Predictive of
Regorafenib Treatment Effect on OS
OS TTP
Regorafenib Regorafenib
Adjusted interaction Adjusted interaction
Protein predictive effect, predictive effect, Referenceǂ
p-value p-value
HR (95% CI)* HR (95% CI)*
1.12 1.10 1 ng/mL
ANG-1 0.019 0.017
(1.05, 1.19) (1.04, 1.17) increase
1.46 1.42 2-fold
Cystatin-B 0.040 0.018
(1.15, 1.85) (1.14, 1.77) increase
1.36 1.41 2-fold
LAP TGF-β1 0.040 0.004
(1.12, 1.65) (1.18, 1.68) increase
1.35 1.78 1 ng/mL
LOX-1 0.009 0.003
(1.16, 1.57) (1.33, 2.39) increase
1.02 1.02 1 pg/mL
MIP-1α 0.040 0.043
(1.01, 1.04) (1.00, 1.03) increase

*An HR >1 indicates enhanced treatment benefit with regorafenib versus placebo with decreased protein levels (or reduced treatment benefit with increased protein levels); ǂReference shows to
what unit increase the HR is related to. CI, confidence interval; HR, hazard ratio; OS, overall survival; TTP, time to progression.
24
Teufel M et al, Gastroenterology 2019
 Lenvatinib vs Sorafenib: Pharmacodynamic Biomarkers
• In biomarker analysis of
REFLECT phase III study
(LEN, n=279; SOR,
n=128), both LEN and
SOR were found to
increase VEGF levels, but
only LEN increased
FGF19 and FGF23 levels,
and decreased ANG-2
levels

*P < 0.05 vs baseline; †P < 0.01 vs baseline ; ‡P < 0.0001 vs baseline; §P < 0.05 between lenvatinib and sorafenib arms.

ANG-2, angiopoietin-2; C#D#, Cycle # Day #; FGF, fibroblast growth factor; LEN, lenvatinib; ORR, objective response rate; OS, overall survival; PFS, progression-free survival;
SOR, sorafenib; TTP, time to progression; uHCC, unresectable hepatocellular carcinoma; VEGF, vascular endothelial growth factor.
25
Finn RS et al ESMO 2018
 Predictive Marker of Response to Sorafenib
Median OS 14 m
EHS No EHS Yes HCV Yes HCV No

NLR ⩽ median NLR > median TB No TB Yes

26
Bruix J et al, J Hepatology 2017
 Median Post Progression-Free Survival According
to the Pattern of Progression in Prospective Studies
Median post progression-free survival, months (95% CI)
Hazard ratio yes/noa or
Trial, study parameters
drug versus placebo
and progression factor
yes or noa Drug Placebo (95% CI)

Reig et al. sorafenib beyond progression

New extrahepatic lesion 7.1 (2.1–12.1) – – 3.01 (1.17–7.76)
New intrahepatic lesion 18.6 (5.8–31.4) – – 1.58 (0.60–4.19)
Intrahepatic growth of existing lesion 14.9 (9.5–20.3) – – 1.30 (0.56–2.99)
Extrahepatic growth of existing lesion 16.7 (4.9–28.4) – – 1.75 (0.63–4.88)
RESORCEb regorafenib versus placebo
New extrahepatic lesion – 9.7 (8.3–12.2) 8.2 (6.1–10.2) 0.70 (0.51–0.97)
New intrahepatic lesion – 12.2 (9.7–15.2) 10.2 (8.8–12.6) 0.79 (0.58–1.08)
Growth and/or extrahepatic of existing
– 12.5 (10.7–14.8) 9.5 (7.8–10.6) 0.64 (0.50–0.81)
intrahepatic lesions
METIV tivantinib versus placebo
New extrahepatic lesion – 6.9 (5–9.5) 8. 2 (6.6–12.1) NA
All patients
–, not –
applicable; NA , not available. aAnalysis performed in patients 8.4 (6.8–10.0)
who were candidates 9.1 (7.3–10.4)
for second- line treatment (n = 43). 0.97 (0.75–1.25)
27 b
Data presented at 26th Conference of the Asian Pacific Association for the Study of the Liver. J Bruix et al, Nat Rev Gastroenterol Hepatol., 2019
 Clinical markers of outcome in sorafenib treatment
Early dermatologic adverse events (DAE60)

Time to progression Overall survival

DAE60
No DAE60
Probability of progression

Probability of survival
p=0.016 p=0.009

DAE60
No DAE60

TTP (months) Survival (months)

Yes: 8.1 months (CI95%:1.6-14.5) Yes :18.2 months (95% CI; 11.9 to 24.4)
vs. vs.
No: 3.9 months (CI95%:2.08-5.7) No:10.1 months (CI 95%:10.1-13.0)
Courtesy of Dr. Maria Reig Reig et al; Journal of Hepatology 2014
 HFSR Correlation with OS in uHCCa
• Subanalysis of RESORCE Trial
100
HFSR (n = 198)
90
Events, n (%), 110 (56)
Probability of survival (%)

80
Median OS (95% Cl), 14.1 months (11.7–16.5)
70
60 No HFSR (n = 181)
50 HFSR Events, n (%), 123 (68)
40
Median OS, 6.6 months (5.0–8.5)

30 HR = 0.52 (95% CI: 0.40–0.68), P<0.0001 (1-sided)

20 No HFSR
10
0
0 3 6 9 12 15 18 21 24 27 30 33

Months from randomization

Number at risk
HFSR 198 179 144 111 82 54 37 25 14 8 5 0
No HFSR 181 138 81 59 41 24 19 9 7 2 0 0
Analysis cutoff date: February 29, 2016.
a

Bruix J et al. Abstract O-009. Presented at: WCGIC 2017; Barcelona, Spain.
29
How to choose and sequence different drugs in HCC?

• Sorafenib: HCV etiology

• Lenvatinib: HBV, improved RR and PFS

• Regorafenib: prior sorafenib tolerability

• Cabozantinib: up to two prior regimens, sorafenib intolerance

• Ramucirumab: baseline AFP > 400 ng/mL

• Nivolumab and pembrolizumab: durable response

 Challenges in Trial Design in HCC

• Surrogate endpoints: PFS/TTP, response rate (RECIST vs

mRECIST)
• Response assessment
• Stratification in HCC trial design
• How to apply biomarker and molecular signature in trial
design

31
 Surrogate Endpoints in Advanced HCC Receiving
Systemic Therapy
• OS is the gold standard in phase III HCC trials
• PFS/TTP?
• Response rate: RECIST vs mRECIST vs (irRECIST)?
• AFP changes?
• Other biomarkers?

32
 Why Considering PFS in Phase III Trials?

• PFS will capture antitumor activity of experimental arm

• OS may be affected by post-progression treatments given
the availability of several approved agents
• Time of completion
• PFS is an accepted primary endpoint in phase III trials in
several tumor types
• The surrogacy of PFS in HCC remains to be established
33
 PFS and OS in Positive Phase III Trials

Phase III trials PFS (months) HR OS (months) HR

Sorafenib vs Placebo 5.5 vs 2.8 0.58 10.7 vs 7.9 0.69 (0.55-0.87)
(SHARP) (TTP)
Sorafenib vs Placebo 2.8 vs 1.4 0.57 6.5 vs 4.2 0.68 (0.50-0.93)
(Asia-Pacific) (TTP)
Lenvatinib vs Sorafenib 7.4 vs 3.7 0.66 13.6 vs 12.3 0.92 (0.79-1.06)
(REFLECT)
Regorafenib vs placebo 3.1 vs 1.5 0.44 10.6 vs 7.8 0.63 (0.50-0.79)
(RESORCE)
Cabozantinib vs 5.2 vs 1.9 0.44 10.2 vs 8.0 0.76 ( 0.63-0.92)
placebo
(CELESTIAL)
Ramucirumab vs 2.8 vs 1.6 0.45 8.5 vs 7.3 0.71 (0.53-0.95)
placebo (REACH-2)

34
 Correlation between PFS/TTP and Overall Survival
Keynote 240: PFS for pembrolizumab vs placebo: 3.0 vs 2.8 months, HR 0.775 (0.609-0.987)

35
Llovet JM, Montal R, Villanueva A. J Hepatol. 2019
 RECIST vs mRECIST

RECIST mRECIST

Widely used clinically and all measurable

Pros Assess viable tumor changes
HCC lesions can be used

More stringent for target lesions selection;

Does not capture tumor necrosis and
Cons
underestimate treatment effect
Extrahepatic lesions and infiltrative lesions
less ideal;
Intra-observer and inter-observer variability;
May be dependent on the mechanism of action of
drugs;
Vasoconstriction related to the drug itself

36
 Studies Assessing the Correlation of mRECIST
Based Responses and OS
• Supportive:
– Brivanib second line phase III trial
– Sorafenib
– Nintedanib
– Lenvatinib
• Lack of correlation:
– Regorafenib in RESORCE phase III trial

37
Llovet et al, JCO 2013; Edeline et al, Cancer 2012; Meyer et al, Liver Int 2007; Bruix et al, Lancet 2017; Kudo et al, GI ASCO 2019
 REFLECT - Global, Randomized, Open-label,
Phase III Noninferiority Study
Lenvatinib Sorafenib Totalendpoint:
Category Primary
Patients with unresectable (n = 478) (n = 476) (n=954)
• OS
HCC (N = 954) Lenvatinib Secondary endpoints:
Median OS, months
• No prior systemic therapy for
Stratification 13.6 12.3
(n = 478) 13.0
• PFS
unresectable HCC • Region: 8 mg (BW <60 kg) or • TTP
95% CI 12.1–14.9 10.4–13.9 11.9-14.1

Randomization 1:1
• ≥1 measurable target lesion per
(Asia-Pacific or 12 mg (BW ≥60 kg) • ORR
mRECIST Western) once daily
• BCLC stage B or C • MPVI and/or EHS: • Quality of life
• Child-Pugh A
• ECOG PS ≤1
ORR , n (%) (Investigator)
†
(yes or no) 115 (24.1) 44 (9.2) 159
• PK(16.7)
lenvatinib exposure
• ECOG PS: parameters
• Adequate organ function
(0 or 1)
95% CI
• Patients with ≥50% liver • Body weight: 20.2–27.9 6.6–11.8
Sorafenib 14.3-19.0
occupation, clear bile duct (<60 kg ≥60 kg) (n = 476) Tumor assessments were
invasion, or portal vein 400 mg twice daily performed according to
ORR, n (%) (Independent)
invasion at the main portal
vein were excluded
194 (40.6) 59 (12.4) mRECIST by the
investigator
95% CI 36.2-45.0 9.4-15.4
BCLC, Barcelona Clinic Liver Cancer; BW; body weight; ECOG PS, Eastern Cooperative Oncology Group Performance Status; EHS, extrahepatic spread; MPVI, macroscopic
Median PFS , months 7.4 3.7
† Response Evaluation Criteria In Solid Tumors; ORR, objective response rate; OS, overall survival; PFS, progression-free survival;
portal vein invasion; mRECIST, modified
PK, pharmacokinetic; RECIST, Response Evaluation Criteria in Solid Tumors; TTP, time to progression.

95% CI 6.9–8.8 3.6–4.6

38
Kudo M et al. Lancet. 2018;391:1163-1173.
 OS by OR for the Overall REFLECT Population
1.0
Landmark KM curve as function of Landmark KM curve as function Median
of OS (months)
Landmark (95% CI) as function of
KM curve
0.9 at 2 months
tumor response tumor response at 4 monthsResponse: 22.4 tumor
(19.7-26.0)
response at 6 months
0.8 Nonresponse: 11.4 (10.3-12.3)
HR (95% CI): 0.61 (0.49-0.76)
0.7 Mantel-Byar Test: P Value: < 0.001
Probability

0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42
Time From Randomization (Month)
Number of patients at risk:
Response 159 155 151 138 121 108 93 76 56 41 22 11 6 1 0
Nonresponse 795 721 571 441 362 291 241 180 129 83 51 26 10 5 0
39 Multivariate analysis showed OR (MPVI, baseline AFP level, treatment etc) correlated with OS
CI, confidence interval; OR, objective response; OS, overall survival.
 Does mRECIST-Based Response Predict OS in
Patients with Advanced HCC?
• Authors conclusion: objective response by mRECIST is an independent predictor of OS in HCC patients in the
REFLECT study
– Large dataset from a positive phase III trial
– Sound statistical methodology
– Potential surrogacy of mRECIST based responses for OS

• Caution:
– Post hoc and retrospective study
– No comparison with RECIST based response analyses
– Not addressing the potential correlation of SD with OS
– Unclear about the correlation between PFS and OS and OR with PFS
– Unclear on the different magnitude of correlation of sorafenib vs lenvatinib
– Additional studies are warranted for prospective validation

40
 mRECIST May Be Dependent on the Mechanism of
Action of Drugs

TKIs Checkpoint inhibitors

(response rate, %) (response rate, %)
RECIST vs Lenvatinib 18.8 vs 40.6 Nivolumab 14 vs 19
mRECIST Nintedanib/ 4.4 vs 15.6 Pembrolizumab 17 vs 15
Sorafenib

41
Kudo et al, Lancet 2018; Meyer et al, Liver Int 2017; El-Khoueiry et al, Lancet 2017; Zhu et al, Lancet Oncol 2018
 Stratification in Phase III Trials
• ECOG PS: subjective
• Macrovascular invasion: extent of vascular invasion
• Extrahepatic disease
• Geographic regions: Asia (except Japan) vs the rest of world
• AFP: 400 ng/mL as cut-off
• Etiology: when sorafenib used as control
• Pattern of progression?

42
 Stratification Used in Phase III Trials

Geographic Viral
ECOG PS MVI EHS AFP
region etiology

SHARP x x x x

RESORCE x x x x x

REFLECT x x x x

CELESTIAL x x x x

REACH-2 x x x

43
 Classification of HCC

44 Nault, Galle, Marquardt J Hepatology 2018

 BLU-554 in Advanced HCC: FGF19 IHC-Positivity
Enriches for Radiographic Tumor Reduction and
Response

*4 confirmed responses

45
Kim et al, ESMO 2017
 REACH Study: Overall Survival in Patients with
Baseline Alpha‑fetoprotein  or < 400 ng/mL

AFP 400 ng/mL AFP <400 ng/mL

46 Zhu AX, et al. Lancet Oncology 2015

 Putative (Adaptive) Design for Future Clinical
Trials in HCC

Umbrella trials
Basket trials

47
Survival According to Response to Nivolumab

Non-Conventional Benefit
• PD with new lesions followed by decrease in target lesion ≥10%
• PD of target lesions followed by decrease in target lesion ≥30%
• PD of target lesions or new lesions followed by stabilization (<10% additional
increase in target lesion + new lesions)

Presented at ASCO GI 2018

Courtesy of Prof. Bruno Sangro

 Predictive Biomarkers to Checkpoint Inhibitors

• MSI status
• PD-L1 expression (cutoff, assays used, expression on
tumor cells, immune cells or both)
• TMB
• Immune signatures

49
Immune Signature of HCC

Sia D et al, Gastroenterology 2017

Pinyol et al, CCR, 2019

50
 Targeted Therapy in Oncogene Addicted Tumors

Crizotinib in ALK-positive NSCLC BGJ398 in FGFR altered CCA

Intrinsic

Acquired

Camidge, et al, Lancet Oncol. 2012 Javle, et al, JCO, 2017

51
 MGH Efforts to Study FGFR Inhibition and Resistance
Clinical Efforts Rapid Autopsy

Breast: 18

Lung: 13 Epithelial: 2

Cholangio: 6 Pancreatic: 2

Colorectal: 3

Melanoma: 2

Laboratory Efforts

/Ğůů>ŝŶĞƐ
WĂƚŝĞŶƚͲ
ĚĞƌŝǀ ĞĚ
yĞŶŽŐƌĂĨƚ;Wy Ϳ
ŝŐĞƐƚΘ 15 FGFR WT
ICC Lines
ƵůƚƵƌĞ
' ĞŶĞƚŝĐĂůůLJͲ
ŶŐŝŶĞĞƌĞĚ 2 FGFR Altered
ICC Lines
ctDNA D ŽƵƐĞD ŽĚĞůƐ;' D D Ɛ Ϳ CTC analysis
ddPCR
52
/K ƌŐĂŶŽŝĚ RNA Seq
 Acquired Resistance to FGFR inhibition in ICC

Overcoming
Acquired resistance FGFR resistance Acquired resistance
to FGFR inhibitors with TAS120 to TAS120

53
 Patient #1: Mutations Detected in Post-progression
Biopsy Tissue
Pre-treatment Nadir (-49.9%) Progression

Tumor biopsy FGFR2-Related Genetic Events Tumor biopsy

Fusion: FGFR2-ZMYMY4 Fusion: FGFR2-ZMYMY4

Mutations: None detected Mutations: FGFR2 V564F
FGFR2 K659M

Plasma (cell-free DNA) Plasma (cell-free DNA)

Mutations: None detected Mutations: FGFR2 N549H

FGFR2 N549K
FGFR2 V564F
FGFR2 E565A
FGFR2 K659M
54
Goyal, Saha, et al, Cancer Discov, 2016
 Patient #2: Response to TAS-120

Mutant Allele Frequency

Patient 73 yo Female with ICC BGJ398 TAS120 FGFR2 (K659M)
1
FGFR2 (K714R)
Mutation FGFR2-SORBS1 fusion
0.1

PFS on BGJ-398 12.4 months

0.01

PFS on TAS-120 15.2+ months Liver biopsy

0.001 showed no
0 10 20 30 FGFR2 mutations
Time (Months)

Response to BGJ398 Response to TAS120

Baseline 8.9 months, nadir Baseline 13.7 months, nadir

-68.2% response -76.5% response

55
Goyal, Shi, Liu et al, Cancer Discov, 2019
 Emerging Targets and Strategies in Advanced HCC
• Novel combinations
– Checkpoint inhibitors and TKIs
– Anti-PD-1/PD-L1+anti-CTLA4
– Anti-PD-L-1+Bevacizumab
• New class of drugs with different mechanism of action
• Novel targets for immunotherapy
• Adoptive Cell based immunotherapy

56
 Durvalumab + Tremelimumab in 1st Line HCC
Phase I/II trial of Durvalumab (PD-L1) + tremelimumab (CTLA-4) demonstrated a 25% ORR 1
(confirmed and unconfirmed) and led to HIMALAYA:

57 1
J Clin Oncol 2017;35 (suppl; abstr 4073)
 Nivolumab and Ipilimumab in advanced HCC

58
Yau T et al, ASCO 2019
Phase Ib Study of Atezolizumab + Bevacizumab in HCC (N=104)

─ Confirmed ORR is 36% per IRF-assessed RECIST 1.1

─ 12% of patients achieved a complete response per IRF-assessed RECIST 1.1
─ PFS 7.3 months, OS 17.1 months
59
Lee et al, APPLE 2019
 Phase 1b Trial: Lenvatinib plus Pembrolizumab
Ikeda et al 2019 AACR Annual Meeting

– Ongoing phase 1b study

– NCT 03006926
– n=30 response-evaluable
– ORR 50% by mRECIST (independent imaging review)
60
– Expansion cohort underway Ikeda et al, AACR 2019
 Conclusions
• Progress has been made in advanced HCC through well conducted
clinical trials
• We have learned many lessens through both negative and positive
trials
• The optimal surrogate endpoints in advanced HCC remain to be
defined
• Biomarker enrichment trials and molecular classification should be
implemented in HCC trials
• New agents and novel strategies are under development

61
 Acknowledgments
MGH
Clinical team:
Medical Oncology: Lipika Goyal, Ryan Corcoran, Dejan
Juric, Jeff Clark, Lawrence Blaszkowsky, Eunice Kwak, Jill
Allen, Jason Faris, Janet Murphy, Dave Ryan
Radiation Oncology: Ted Hong, Jen Wo, Tom DeLaney
Surgery/Transplant: Ken Tanabe, Christina Ferrone, Dave
Berger, Carlos Fernendez-del Castillo, Keith Lillemore, Parsia
Vagefi, Nahel Elias, James Markmann
Steele Lab: Dan Duda, Rakesh Jain, Becky Chen, Tai Hato,
Translational Research Laboratory (TRL): Johan Iafrate,
Darrell Borger, Jochen Lennerz
Bardeesy Lab: Leah Liu, Christine Parachoniak, Nabeel
Bardeesy
Corcoran Lab: Ryan Corcoran
Phase I: Dejan Juric, Keith Flaherty
Bardelli Lab: A Bardelli
Novartis Institute for BioMedical Research: Ralph Tiedt,
Diana Graus Porta
Research Assistants: Stephanie Reyes, Laura Henderson,
Emily E Van Seventer
DFCI
Tom Abrams, James Cleary, Matthew Kulke, Peter
Enzinger, Pankaj Bhargava, Jeffrey Meyerhardt,
Jennifer Chan, Brian Wolpin, Kimmie Ng, Robert
Mayer
Yale: Charles Fuchs
BIDMC:
Andrea Bullock
U. of Rochester: Aram Hezel
U. of Washington/Fred Hutchinson: Supriya Saha
Ohio State University: Tim Pawlik
Medical College of Wisconsin: T. Clark Gamblin
The University of Hong Kong: Thomas Yau,
Ronnie Poon
Funding support for this project:
Cholangiocarcinoma Foundation, NIH GI SPORE, V
Foundation Translational Grant, Jonathan Kraft
Translational Research Award
ILCA and many colleagues and friends in the
HCC/cholangiocarcinoma community