Association of Nucleos(t)ide Analogue Therapy With Reduced Risk of Hepatocellular Carcinoma in Patients With Chronic Hepatitis B—A Nationwide Cohort Study

All studies published in Gastroenterology are embargoed until 3PM ET of the day they are published as corrected proofs on-line. Studies cannot be publicized as accepted manuscripts or uncorrected proofs.

Gastroenterology 2014;-:1–9

Association of Nucleos(t)ide Analogue Therapy With Reduced Risk of Hepatocellular Carcinoma in Patients With Chronic Hepatitis B—A Nationwide Cohort Study Q15

Chun-Ying Wu,1,2,3,4 Jaw-Town Lin,5,6,7 Hsiu J. Ho,5 Chien-Wei Su,2,8,9 Teng-Yu Lee,2,3 Shen-Yung Wang,8 Chuhui Wu,8 and Jaw-Ching Wu8,10 1

Division of Gastroenterology, Taichung Veterans General Hospital, Taichung, Taiwan; 2Faculty of Medicine, School of Medicine, National Yang-Ming University, Taipei, Taiwan; 3Department of Public Health and Graduate Institute of Clinical Medicine, China Medical University, Taichung, Taiwan; 4Department of Life Sciences, National Chung-Hsing University, Taichung, Taiwan; 5School of Medicine, Fu Jen Catholic University, Taipei, Taiwan; 6Department of Internal Medicine, E-Da Hospital/I-Shou University, Kaohsiung, Taiwan; 7Institute of Population Health Sciences, National Health Research Institutes, Miaoli, Taiwan; 8Institute of Clinical Medicine and Genomic Research Center, National Yang-Ming University, Taipei, Taiwan; 9Division of Gastroenterology, Taipei Veterans General Hospital, Taipei, Taiwan; and 10Translational Research Division, Medical Research Department, Taipei Veterans General Hospital, Taipei, Taiwan BACKGROUND & AIMS: Treatment for hepatitis B virus infection reduces the risk of hepatocellular carcinoma (HCC). However, the long-term protective effects for subgroups of patients with chronic hepatitis B are unclear. METHODS: We conducted a retrospective nationwide cohort study using data from Taiwan’s National Health Insurance Research Database (from January 1, 1997, through December 31, 2010). Cumulative incidences were calculated and multivariable analyses were carried out after adjusting for competing mortality. Propensity scores were used to match 21,595 patients with chronic hepatitis B who received nucleoside analogue therapy for at least 90 days (treated cohort) with 21,595 untreated patients with chronic hepatitis B (controls), who received hepatoprotectants for at least 90 days. Data were collected from the treated cohort for a mean period of 3.46 years and from controls for 5.24 years. RESULTS: The treated cohort had a significantly lower 7-year incidence of HCC (7.32%; 95% confidence interval [CI], 6.77%
7.87%) than controls (22.7%; 95% CI, 22.1%
23.3%; P < .001). After adjusting for competing mortality and other confounders, nucleos(t)ide analogue treatment was associated with a reduced risk of HCC, with an adjusted hazard ratio of 0.37 (95% CI, 0.34–0.39; P < .001). Sensitivity analyses confirmed the association between nucleos(t)ide analogue treatment and reduced risk of HCC. Age, sex, cirrhosis, and diabetes mellitus modified this association. CONCLUSIONS: Based on a retrospective, nationwide study in Taiwan, nucleoside analogue therapy use is associated with reduced risk of HCC in patients with chronic hepatitis B virus infection.

hepatitis C virus, alcoholic liver disease, nonalcoholic steatohepatitis, and metabolic syndrome.3 Globally, and in Asia particularly, chronic HBV infection is the most frequent underlying cause of HCC and accounts for approximately half of HCC cases.4,5 Vaccines against HBV have successfully reduced the incidence of HBV in the younger generations, however, there are still >350 million patients infected with HBV worldwide.6 Chronic hepatitis B (CHB) infection not only causes hepatitis, but also leads to hepatic decompensation, cirrhosis, and hepatocellular carcinoma (HCC).1,3,6,7 HBV replication has been identified as a major element of immune-mediated liver tissue injury and disease progression. Higher HBV DNA levels are associated with increased risk of HCC development and recurrence.4,8,9 Nucleos(t)ide analogue therapy effectively suppresses HBV replication by inhibiting HBV polymerase.10,11 Treatment with nucleos(t)ide analogues has been reported to delay disease progression in CHB patients.12,13 Regression of liver cirrhosis has been observed with long-term use of nucleos(t)ide analogues.14,15 In addition, nucleos(t)ide analogue therapy has been reported to be associated with reduced risk of HCC development and recurrence.16–18 In a meta-analysis that pooled 5 studies and a total of 2289 CHB patients, the risk of HCC was reduced by 78% among those receiving nucleos(t)ide analogue therapy relative to controls that did not received nucleos(t)ide analogue.19 In another systematic review of 21 studies involving 3881 treated and 534 untreated patients, the protective roles of nucleos(t)ide analogues in the development of HCC were

Keywords: Hepatoma; HBV; Antiviral Agent; NHIRD.

H

epatocellular carcinoma (HCC) is the fifth most common cancer worldwide and the third most common cause of cancer-related mortality. It is estimated that >748,000 new HCC cases and about 700,000 deaths occur annually worldwide.1,2 Many risk factors contribute to the development of HCC, including hepatitis B virus (HBV),

Abbreviations used in this paper: CHB, chronic hepatitis B; CI, confidence interval; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HR, hazard ratio; ICD-9, International Classification of Diseases, 9th Revision; NHI, National Health Insurance; NHIRD, National Health Insurance Research Database; NNT, number needed to treat. © 2014 by the AGA Institute 0016-5085/$36.00 http://dx.doi.org/10.1053/j.gastro.2014.03.048

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 CLINICAL LIVER

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59

2 Q5

CLINICAL LIVER

119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 Q6 165 166 167 168 169 170 171 172 173 174 175 176 177

Wu et al

Gastroenterology Vol.

confirmed.20 In the CALM study, nucleos(t)ide analogue therapy attenuated 51% risk of HCC in patients with CHB.12 However, some literature report no beneficial effects for nucleos(t)ide analogue in reducing HCC risk among CHB patients. In a recent cohort study, the risk of HCC development in patients with oral antiviral therapy is still significantly higher than in patients with inactive CHB.21 In another study, oral nucleos(t)ide analogue reduces the incidence of cirrhosis and risk of complications, but not development of HCC in cirrhotic patients.22 With this controversial evidence, a long-term population-based nationwide study will be helpful to investigate the association between nucleos(t)ide analogue treatment and risk of HCC in HBV-infected patients. In the present study, we conducted a nationwide cohort study to examine whether nucleos(t)ide analogue use is associated with reduced risk of HCC in CHB patients. In addition, we examined the number needed to treat (NNT) for 1 less HCC development.

Methods Study Design We conducted this nationwide cohort study based on Taiwan’s National Health Insurance Research Database (NHIRD). The NHIRD was set up in 1997 when the National Health Insurance (NHI) program, a compulsory universal health insurance program for nearly all 23.7 million residents in Taiwan, was established. Comprehensive health care information, including diagnoses, prescriptions, and laboratory check-up items can be retrieved from the NHIRD, which has been described in detail in our previous studies.18,23,24 International Classification of Diseases, 9th Revision (ICD-9) codes are used to define diseases in this database. The accuracy of diagnosis of major diseases, such as ischemic stroke and acute coronary syndrome, has been validated.25,26 This study has been approved by the ethical review board of the National Health Research Institutes, Taiwan.

Study Subjects CHB patients in the NHIRD were defined as those meeting the following 2 criteria: diagnosed with CHB (ICD-9 codes: 070.2, 070.3, and V02.61) 3 times in outpatient clinic or admitted with a diagnosis of HBV infection between January 1, 1997 and December 31, 2010 and having used nucleos(t)ide analogues or hepatoprotective agents (eg, silymarin, liver hydrolysate, and choline bitartrate). These CHB patients were followed up from the time of diagnosis of CHB until development of HCC, death, or December 31, 2010. We first excluded patients not using nucleos(t)ide analogues or hepatoprotective agents for at least 90 days. Nucleos(t)ide analogues have been covered under the NHI program for CHB patients since October 1, 2003. However, reimbursement for nucleos(t)ide analogues requires patients to fulfill certain criteria, such as twice-elevated serum alanine aminotransferase ([ALT] 2) and elevated HBV DNA titer (>2000 IU/mL). The reimbursement duration ranges from 18 months to 36 months.18 The reimbursement for hepatoprotective agents requires only elevated ALT level

-,

No.

-

(ie, ALT 1). Hepatoprotective agents have been reimbursed since the beginning of Taiwan’s NHI program in 1997. Next, we excluded patients with hepatitis C virus (ICD-9 codes: 070.41, 070.44, 070.51, 070.54, and V02.62), human immunodeficiency virus (ICD-9 code: 042), other viral hepatitis (ICD-9 code: V02.69), and malignant tumors (ICD-9 codes: 104–208).

Study Cohorts Among the eligible patients, the treated patients were defined as those who received nucleos(t)ide analogues for at least 90 days. Those who did not use nucleos(t)ide analogues for at least 90 days were defined as nontreated patients. Demographic data of these 2 groups were first compared. Then, propensity scores for estimating the probability of receiving nucleos(t)ide analogue therapy were developed using the logistic regression model to estimate the differences in baseline characteristics between the treated patients and the untreated patients. The propensity score approach used in our study was described in Dehejia et al.27 We matched each treated patient with 1 untreated patient, based on propensity scores. Histograms before and after matching were examined to assess the success of the propensity scores in balancing the 2 groups. The index date of follow-up was the first date of nucleos(t)ide analogue prescription for the treated cohort and the first date of hepatoprotective agent prescription for the untreated cohort.

Hepatocellular Carcinoma Risk Analysis HCC diagnosis was defined according to the major diagnosis of admission (ICD-9 codes: 155.0 and 155.2) or enrollment in the Registry for Catastrophic Illness Patient Database, a subset of the NHIRD. Patients were registered in the Registry for Catastrophic Illness Patient Database if their diagnoses were confirmed by pathology reports or typical imaging presentations. The first date of admission or enrollment in the Registry for Catastrophic Illness Patient Database was defined as the date of HCC development. We excluded patients with a diagnosis of HCC in the first 90 days after start of nucleos(t)ide analogue therapy or hepatoprotective agents. Cumulative incidences of HCC were analyzed after adjusting for competing mortality. Because death usually results from underlying comorbidities that can also impact HCC risk, its occurrence leads to informative censoring in calculating HCC incidence. Therefore, death before HCC development was considered a competing risk event on survival analysis. Death-adjusted cumulative incidences in competing risk data ratios were analyzed using modified Kaplan-Meier method and Gray method.28,29 The R package “cmprsk” (http://cran.r-project. org/web/packages/cmprsk/index.html) was used in the competing risk analyses.

Multivariable Analyses To determine whether nucleos(t)ide analogue use was independently associated with reduced risk of HCC, the Cox proportional hazards model was developed. Because the case numbers in the present study were large enough, all

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236

237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295

2014

HBV Therapy and HCC Incidence

the parameters defined a priori were used in the model to exclude as many confounders as possible. To examine potential heterogeneity of treatment effect in relation to confounders, we performed interaction analyses by adding interaction terms between treatment and potential confounding factors, including age, sex, cirrhosis, liver decompensation, comorbidities, use of statins, use of nonsteroidal anti-inflammatory drugs, and use of metformin in the multivariable analyses. Because the sample size is quite large in the present study, we defined the a level at .05 as significantly different for the interaction instead of using an a level at .10. Once a factor is found significantly interacted with treatment, multivariable stratified analyses were conducted to examine the associations of nucleos(t)ide use and risk of HCC in CHB patients with these factors.

Sensitivity Analyses Nucleos(t)ide analogues have been covered under the NHI program since October 1, 2003, midway through the follow-up period. Therefore, the follow-up durations in the 2 treatment groups differed significantly. The impact of reimbursement of antivirals in 2003 can induce several other unmeasured factors, such as different screening policy, different accessibility of health care, and different health consciousness of patients, etc. To address the differential follow-up in the 2 treatment groups, we conducted sensitivity analyses with fixed duration by limiting the index date of follow-up to between October 1, 2003 and September 30, 2005 and the follow-up duration to 5 years. To examine the impact of other potential unmeasured confounders on the estimated treatment effect, we performed sensitivity analyses with add-on of an unmeasured confounder according to the method of Lin et al30 using the R package “obsSens.” In the sensitivity analyses, we added another hypothetical unmeasured confounder with a similar favorable protective effect as our antiviral agents. Then we examined how this added unmeasured factor confounded our observations with different prevalences in the treated and untreated groups.

3

at least 90 days. Patients with hepatitis C, other hepatitis, 296 human immunodeficiency virus, and those with malig- 297 nancies before the use of nucleos(t)ide analogues or hep- 298 atoprotectants were excluded. Among the remaining 72,458 299 patients, 47,611 patients were in the untreated group and 300 24,847 patients were in the treated group. These 2 groups 301 had significant differences in their demographic data 302 (Supplementary Table 1). To estimate the probability of 303 receiving nucleos(t)ide analogue therapy, age, sex, comor- 304 bidities (in separate terms for individual comorbidities), use 305 of statins, use of nonsteroidal anti-inflammatory drugs, 306 and use of metformin were used to calculate propensity 307 scores. The original propensity scores of the untreated pa- 308 tients (mean, 0.29) were significantly lower when compared 309 with the treated cohort (mean, 0.44) (P < .001). We used 310 propensity score to match 1 patient in the treated cohort 311 with 1 patient in the untreated cohort. The histograms of 312 propensity score before and after matching are shown in 313 Supplementary Figure 1. Before matching, the untreated 314 patients had significantly lower propensity scores (black 315 bars in the histogram) compared with the treated patients 316 (gray bars in the histogram) (Supplementary Figure 1A). 317 After matching, the 2 groups had comparable distributions 318 of propensity scores (Supplementary Figure 1B). Finally, we 319 recruited 21,595 patients into the treated cohort and 21,595 320 patients into the untreated cohort (Figure 1). In the treated Q7321 CLINICAL LIVER

-

Statistical Analysis The demographic characteristics of the treated and untreated patients were compared using the c2 test and Student t test. The number of patients who needed to be treated for 1 less HCC occurrence was defined as NNT and calculated by the inverse of the absolute risk reduction. The confidence intervals of NNT were calculated according to the method of Altman et al in a survival setting.31 All data management was performed using SAS 9.1 software (SAS Institute, Cary, NC). Cumulative incidences were analyzed using the survival and EpiTools packages of R.

Results Demographic Data Between 1997 and 2010, a total of 199,451 patients were diagnosed with CHB. Among them, 81,823 patients had used nucleos(t)ide analogues or hepatoprotectants for

Figure 1. Study patient selection flow diagram. *A case can be excluded due to more than one criterion. Therefore, the total excluded cases in each step can outnumber the sum of case numbers excluded by individual criteria. HIV, human immunodeficiency virus.

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354

4

CLINICAL LIVER

355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413

Wu et al

Gastroenterology Vol.

cohort, a total of 19,063 patients received only 1 nucleos(t) ide analogue, including 12,938 patients who received lamivudine, 5748 patients who received entecavir, and 377 patients who received telbivudine. The remaining 2532 patients received more than 1 nucleos(t)ide analogue.

-,

No.

-

As shown in Table 1, the mean age of these 2 cohorts was 43.5 years and about three fourths of the patients were male. Mean follow-up durations for the treated and untreated cohorts were 3.46 years and 5.24 years, respectively. Mean sonography and a-fetoprotein screening

Table 1.Baseline Demographic Characteristics and Outcomes of Study Cohorts Demographic characteristics and outcomes Age, y, mean  SDd Sex, n (%) Female Male Follow-up yearse Mean  SD Median (IQR) Sonography screening (n/year) Mean  SD Median (IQR) AFP screening (n/year) Mean  SD Median (IQR) Nucleos(t)ide analogue therapy duration, y Mean  SD Median (IQR) Hepatoprotective agents, y Mean  SD Median (IQR) Concomitant drug users,f n (%) Statin NSAIDs or aspirin Metformin Major coexisting diseases, n (%) Cirrhosis Liver decompensation Hypertension Diabetes COPD Acute coronary syndrome Cerebral vascular disease Renal failure Hypercholesterolemia Charlson’s score Mean  SD Median (IQR) Propensity scoreg Mean  SD Median (IQR) Events, n (%) HCC occurrence Death before HCC occurrence Overall death

Untreateda (n ¼ 21,595)

Treatedb (n ¼ 21,595)

43.52  13.38

43.53  13.61

4982 (23.1) 16,613 (76.9)

P valuec .935 .001

5281 (24.5) 16,314 (75.5)

5.24  2.17 6.51 (3.54–7.00)

3.46  2.2 3.34 (1.40–5.50)

<.001 <.001

1.90  1.58 1.54 (0.83–2.48)

2.08  1.61 1.85 (1.16–2.63)

<.001 <.001

1.81  1.69 1.38 (0.56–2.55)

2.88  2.29 2.41 (1.38–3.84)

<.001 <.001

1.44  0.74 1.42 (1.02–1.68) 1.24  1.28 0.77 (0.43–1.55) 1413 (6.5) 11,996 (55.5) 1880 (8.7) 3016 1646 1827 1574 486 674 506 389 206

(14.0) (7.6) (8.5) (7.3) (2.3) (3.1) (2.3) (1.8) (1.0)

0.78  1.14 0.36 (0.08–0.98) 1474 (6.8) 11,903 (55.1) 2000 (9.3) 2847 1695 1893 1590 462 654 499 414 201

(13.2) (7.8) (8.8) (7.4) (2.1) (3.0) (2.3) (1.9) (0.9)

<.001 <.001 .248 .373 .045 .018 .387 .265 .782 .450 .596 .848 .393 .842

0.80  1.53 0 (0–1)

0.79  1.52 0 (0–1)

.308 .165

0.42  0.16 0.42 (0.3–0.53)

0.42  0.16 0.42 (0.3–0.53)

.476 .397

4454 (20.6) 2556 (11.8) 4778 (22.1)

992 (4.6) 1036 (4.8) 1406 (6.5)

<.001 <.001 <.001

AFP, a-fetoprotein; COPD, chronic obstructive pulmonary disease; HCC, hepatocellular carcinoma; IQR, interquartile range; NSAIDs, nonsteroidal anti-inflammatory drugs. a Untreated, not receiving nucleos(t)ide analogues. b Treated, receiving nucleos(t)ide analogues. c P values were compared using c2 test and Student t test. d Age is treated as a continuous variable. e Follow-up is defined as the time of nucleos(t)ide analogue or hepatoprotective treatment. f Drug users indicate patients using drugs at least 1 day per month on average. g Age, sex, acute coronary syndrome, cerebral vascular diseases, COPD, diabetes, cirrhosis, HCC, liver decompensation, renal failure, hypertension, hypercholesterolemia, use of statins, use of NSAIDs or aspirin or coxibs, and use of metformin were included in the propensity score calculation.

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472

473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531

2014

HBV Therapy and HCC Incidence

frequencies (number per year) were 2.08 and 2.88 for the treated cohort and 1.90 and 1.81 for the untreated cohorts, respectively. Mean duration of nucleos(t)ide analogue use in the treated cohort was 1.44 years. Mean duration of hepatoprotectant use in the untreated cohort was 1.24 years. The treated cohort had a significantly lower incidence of HCC (n ¼ 992 [4.6%]) when compared with the untreated cohort (n ¼ 4454 [20.6%]). Competing mortality (death before development of HCC) was significantly lower in the treated cohort (n ¼ 1036 [4.8%]) than in the untreated cohort (n ¼ 2256 [11.8%]). Overall mortality in the treated cohort (n ¼ 1406 [6.5%]) was also significantly lower than in the untreated cohort (n ¼ 4778 [22.1%]).

Seven-Year Cumulative Incidences of Hepatocellular Carcinoma for Treated and Untreated Cohorts Cumulative incidences of HCC after adjustment for competing mortality are shown in Figure 2. Patients in the treated cohort were associated with a significantly lower risk of HCC (7-year cumulative incidence: 7.32%; 95% confidence interval [CI]: 6.77%
7.87%) than those in the untreated cohort (22.70%; 95% CI: 22.11%
23.30%) (P < .001). On average, the annual incidences of HCC in treated and untreated cohorts were 1.05% and 3.24%, respectively. The unadjusted NNT associated with 1 less HCC development within 7 years was 7 (95% CI: 6.2–6.9). This suggests that use of nucleos(t)ide analogues in 7 CHB patients is associated with 1 less HCC development within 7 years.

Multivariable Analysis Without controlling for other factors, nucleos(t)ide analogue treatment was associated with reduced risk of HCC development (hazard ratio [HR] ¼ 0.34; 95% CI: 0.32–0.37;

P < .001). After adjusting for competing mortality and other confounders, we found that nucleos(t)ide analogues treatment is associated with a significantly lower risk of HCC (HR ¼ 0.37; 95% CI: 0.34–0.39; P < .001). Older age, male sex, and liver cirrhosis were found to be risk factors for increased HCC risk. Patients with comorbidities including liver decompensation, hypertension, chronic obstructive pulmonary disease, acute coronary syndrome, and cerebral vascular diseases were found to be associated with reduced risk of HCC due to higher competing mortality. Use of statin and use of nonsteroidal anti-inflammatory drugs or aspirin were associated with significantly lower risk of HCC in CHB patients (Table 2). To examine whether significant heterogeneity of treatment effect exists in relation to age, sex, cirrhosis, liver decompensation, diabetes, and other potential confounders, we added interaction terms to the multivariable analyses (Supplementary Table 2). On the interaction analysis, we found statistically significant interactions between nucleos(t)ide use and age, sex, liver cirrhosis, and diabetes. Because the interactions are statistically significant, we cannot interpret the main effect of treatment. Instead, we examined the effect of treatment within each level of the factors, including age, sex, liver cirrhosis, and diabetes. In Figure 3, we conducted multivariable subgroup analyses. We found that the treated cohort was associated with a reduced risk of HCC in all subgroups. The beneficial effect of nucleoside analogues was especially significant among younger patients (younger than 40 years old: HR ¼ 0.13;

Table 2.Multivariable Cox Proportional Hazards Model Analysis for Risk of Hepatocellular Carcinoma Occurrence After Adjusting for Competing Mortality

Treated vs untreated Age, per each incremental year Male Cirrhosis Liver decompensation Hypertension Diabetes COPD ACS CVA Renal failure Hypercholesterolemia Statin use NSAIDs or aspirin use Metformin use

Figure 2. Cumulative incidences of HCC after adjustment for competing mortality. Calculation and comparison of cumulative incidences in competing risk data ratios were conducted using modified Kaplan–Meier method and Gray’s method. Patients who developed HCC during the first 3 months were excluded. Untreated, CHB patients not receiving nucleos(t)ide analogues; Treated, CHB patients receiving nucleos(t)ide analogues.

5

Hazard ratioa

95% CI

P value

0.37 1.06

0.34–0.39 1.06–1.06

<.001 <.001

1.52 1.92 0.87 0.81 1.05 0.71 0.66 0.50 0.84 0.87 0.55 0.62 0.97

1.42–1.63 1.77–2.08 0.78–0.97 0.72–0.90 0.93–1.17 0.59–0.87 0.55–0.79 0.40–0.64 0.68–1.04 0.61–1.23 0.47–0.63 0.58–0.65 0.879–1.073

<.001 <.001 .013 <.001 .450 .001 <.001 <.001 .110 .430 <.001 <.001 .570

ACS, acute coronary syndrome; CI, confidence interval; COPD, chronic obstructive pulmonary disease; CVA, cerebral vascular diseases; NSAID, nonsteroidal anti-inflammatory drug. a Adjusted for covariate factors, including age, sex, comorbidities, use of statins, use of NSAIDs or aspirin, and use of metformin.

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 CLINICAL LIVER

-

6

Gastroenterology Vol.

CLINICAL LIVER

591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 Q8 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649

Wu et al

40–50 years old: HR ¼ 0.30; 50 years or older: HR ¼ 0.49), patients without cirrhosis (noncirrhosis vs cirrhosis: HR ¼ 0.27; vs HR ¼ 0.72), and patients without diabetes (nondiabetes vs diabetes: HR ¼ 0.34 vs HR ¼ 0.69).

Sensitivity Analysis In the sensitivity analysis with fixed duration, we only identified CHB patients between October 1, 2003 and September 30, 2005 because of 3 reasons. First, nucleos(t) ide analogues were covered under the NHI program since October 1, 2003. Second, we wanted to follow these patients up to 5 years, until the end of 2010. Third, these patients much used nucleos(t)ide analogues or hepatoprotectants for at least 90 days. Propensity scores were used to match each treated patient with 1 untreated patient. Finally, we identified 4545 patients in the treated cohort and 4545 patients in the untreated cohort. The demographic characteristics and outcomes are shown in Supplementary Table 3. Patients in the treated cohort were associated with significantly lower risk of HCC development (5-year cumulative incidence: 6.62%; 95% CI: 5.90%
7.35%) than those in the untreated cohort (19.08%; 95% CI: 17.93%
20.22%) (P < .001) (Supplementary Figure 2). On multivariable analysis, the treated cohort was associated with reduced risk of HCC development (adjusted HR ¼ 0.31; 95% CI: 0.27–0.53; P < .001) (Supplementary Table 4). In Supplementary Figure 3, we used sensitivity analysis to examine the trend of estimates of the treated hazard on covariate-adjusted Cox model with add-on of an unmeasured confounder with relative hazard of 0.3. When all the subjects in untreated group have the add-on unmeasured confounder (prevalence of the confounder in the untreated group is 1.0) and none of subjects in treated group has this

-,

No.

-

650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 Figure 3. Multivariable 665 stratified analyses. Among chronic CHB patients, 666 nucleoside analogue use 667 (treated cohort) is associ- 668 ated with reduced risk of 669 HCC development in all 670 subgroups. All P values 671 were significant. 672 673 674 unmeasured confounder (prevalence of the confounder in 675 the treated group is 0.0), then the impact of antiviral ther- Q9676 apy would be beneficial (HR ¼ 0.1, the bottom line in 677 Supplementary Figure 3). On the contrary, when none of 678 subjects in the untreated group has the add-on unmeasured 679 confounder (P0 ¼.0) and all subjects in treated group have Q10680 this confounder; then the impact of antiviral therapy would 681 not be protective (HR ¼ 1.2, the top line in Figure 3). In most 682 situations, patients who received nucleos(t)ide analogues 683 had lower risk of HCC occurrence relative to those who did 684 not, even if a favorable unmeasured confounder exists. 685 686 687 Discussion 688 This population-based cohort study demonstrated that 689 use of nucleos(t)ide analogues is associated with reduced 690 long-term risk of HCC in CHB patients. After adjusting for 691 death as the competing cause of risk and for multiple con- 692 founding factors, we found that use of nucleos(t)ide ana- 693 logues is associated with an adjusted HR of 0.37 for HCC 694 occurrence in CHB patients. The association between 695 nucleos(t)ide analogues use and lower risk of HCC was found 696 in all subgroups of CHB patients, especially in younger pa- 697 tients, patients without liver cirrhosis, patients without 698 liver decompensation, and patients without diabetes. We 699 also validated our observations by sensitivity analyses. 700 In the present study, the association between use of 701 nucleos(t)ide analogues and risk of HCC in CHB patients 702 diminished with age. The HRs associated with use of 703 nucleos(t)ide analogues were 0.13, 0.30, and 0.49 for pa- 704 tients younger than 40 years old, aged between age 40 and 705 50 years, and older than 50 years, respectively. Several 706 reasons might explain this observation. First, this interac- 707 tion resulted from the rising probability of HCC in CHB 708

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767

Q11

2014

HBV Therapy and HCC Incidence

patients with advanced age. In the REVEAL study, which investigated the natural history of CHB patients in Taiwan, Chen et al found that the risk of HCC in CHB patients remains low before age 40 years, starts to rise in the 40s, and significantly increases after age 50 years.7 The rising probability of HCC with advanced age might be reflected in the absolute risk reduction by use of nucleos(t)ide analogues in different age groups. The 7-year HCC absolute risk reductions for patients younger than 40 years, aged 40–50 years, and older than 50 years were 7.92%, 18.23%, and 18.68%, respectively. Second, older patients might have more comorbidities, which leads to higher risk of competing mortality. In the present study, we found that patients with hypertension, chronic obstructive pulmonary disease, acute coronary syndrome, and cerebral vascular diseases have higher competing mortality and lower HCC risk after adjusting for competing mortality. Third, the starting time for nucleos(t)ide analogue use might be too late to rescue the carcinogenesis of HCC. However, we need more evidence to support this hypothesis. For noncirrhotic CHB patients, mean annual incidences of HCC were 0.68% and 2.97% for the treated and the untreated cohorts, respectively. Adjusted HR was 0.25 for the use of nucleos(t)ide analogues. Our observations of the treated group were comparable with those of previous reports, but the annual HCC incidence in the untreated noncirrhotic patients in the present study was higher than in previous studies.13,20,32 In a meta-analysis based on 5 studies comparing patients treated with nucleos(t)ide analogues with controls, Sung et al reported that the risk of HCC after nucleos(t)ide analogue treatment is reduced by 78%.19 In another systematic review of 21 studies by Papatheodoridis et al, HCC developed in 2.8% and 6.4% of nucleos(t) ide analogue-treated and untreated CHB patients, respectively, during a 46-month period.20 In a retrospective cohort study based on 377 CHB patients (17% with cirrhosis), annual HCC incidences were 0.4% for nucleos(t)ide analogue-treated group and 2.5% for control group, respectively.33 The higher annual HCC incidence in our untreated noncirrhotic patients might be due to the requirement for higher baseline aminotransferase levels to obtain reimbursement for hepatoprotective agents. For cirrhotic CHB patients, the mean annual incidences of HCC in the present study were 3.90% and 4.94% for treated and untreated cohorts, respectively. The adjusted HR was 0.72 for the use of nucleos(t)ide analogue. The annual incidence of our treated cohort was slightly higher than that of previous studies. In a retrospective cohort study based on CHB patients with cirrhosis, the annual incidences of HCC in nucleos(t)ide analogue treated and untreated groups were 1.02% and 6.0%, respectively.34 In the CALM study, a randomized trial based on CHB patients with cirrhosis or advanced fibrosis, nucleos(t)ide analogue use was found to reduce risk of HCC development during a median duration of 32.4 months of therapy (3.9% vs 7.4%; P ¼ .047).12 A possible explanation for the higher annual incidences in the treated cohort in the present study is the strict NHI regulations regarding nucleos(t)ide analogue reimbursement.18 Only high-risk populations, including

7

patients with higher baseline HBV viral load, higher ALT level, or higher prevalence of liver decompensation are eligible for reimbursement. These higher-risk populations can contribute to the higher annual incidences. The chemopreventive effect of nucleos(t)ide analogue therapy in the present study was significantly higher in nondiabetic patients when compared with diabetic patients (adjusted HR ¼ 0.34 vs 0.69). In our recent nationwide casecontrol study, we found that diabetes is independently associated with increased risk of HCC development (odds ratio ¼ 2.25).35 In the United States, diabetes has also been found to be associated with 2- to 3-fold increase in the risk of HCC, regardless of other HCC risk factors, such as viral hepatitis.36,37 Several factors can explain the association between HCC and diabetes, such as increased risk for nonalcoholic fatty liver disease and nonalcoholic steatohepatitis in diabetic patients. Metformin has been found to be associated with a decreased risk of HCC in diabetes patients, via inhibition of hepatoma cell proliferation and induction of cell cycle arrest at G0/G1 phase.35 In the era of nucleos(t)ide analogue therapy, more studies are needed to investigate the role of diabetes in the carcinogenesis of HCC to further decrease HCC incidence. Recently, several HCC score calculators have been introduced.38–40 Unfortunately, we did not have information on baseline ALT and HBV DNA levels, which have been shown to be associated with HCC risk in these HCC score calculators. Given that reimbursement for nucleos(t)ide analogues requires twice-elevated aminotransferase and higher HBV DNA levels (>2000 IU/mL), and reimbursement for hepatoprotective agents (control group) requires only elevated aminotransferase level (ALT 1), the higher baseline HBV DNA levels and ALT levels in the treated cohort might have led to a more conservative estimation of the protective effect of nucleos(t)ide analogues. In the present study, we used many methods to prevent potential confounders. However, some unmeasurable bias can still exist. Propensity score matching was used to select comparable controls to imitate a randomized clinical trial. Although we used all potential confounders in the model to create propensity score, there were some significant differences in the distributions, such as for sex and liver cirrhosis. The large sample size in the present study might be the reason for the statistical significance. For examples, the differences between 24.5% female in the treated cohort vs 23.1% female in the untreated cohort, and the differences of 13.2% vs 14.0% cirrhosis in the treated and untreated cohorts are statistically significant, but might be not clinically significant. Reimbursement for nucleos(t)ide analogues began in 2003, midway through the follow-up period, which might also have confounded our observations. However, protective role of nucleos(t)ide analogue uses in HCC risk was found on sensitivity analyses with fixed duration. In conclusion, nucleos(t)ide analogue use is associated with reduced risk of HCC in CHB patients. Age, sex, liver cirrhosis, and diabetes mellitus modify this association. More studies are needed to explore the wider use of nucleos(t)ide analogues for prolonged periods to decrease the incidences of HCC.

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 CLINICAL LIVER

-

8

CLINICAL LIVER

827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 Q13 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885

Wu et al

Gastroenterology Vol.

Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at http://dx.doi.org/10.1053/ j.gastro.2014.03.048.

19.

20.

References 1. Jemal A, Bray F, Center MM, et al. Global cancer statistics. CA Cancer J Clin 2011;61:69–90. 2. El-Serag HB, Rudolph KL. Hepatocellular carcinoma: epidemiology and molecular carcinogenesis. Gastroenterology 2007;132:2557–2576. 3. El-Serag HB. Hepatocellular carcinoma. N Engl J Med 2011;365:1118–1127. 4. Liu CJ, Kao JH. Hepatitis B virus-related hepatocellular carcinoma: epidemiology and pathogenic role of viral factors. J Chin Med Assoc 2007;70:141–145. 5. Sherman M, Llovet JM. Smoking, hepatitis B virus infection, and development of hepatocellular carcinoma. J Natl Cancer Inst 2011;103:1642–1643. 6. Liaw YF, Chu CM. Hepatitis B virus infection. Lancet 2009;373:582–592. 7. Chen CJ, Yang HI. Natural history of chronic hepatitis B REVEALed. J Gastroenterol Hepatol 2011;26:628–638. 8. Chen CJ, Yang HI, Su J, et al. Risk of hepatocellular carcinoma across a biological gradient of serum hepatitis B virus DNA level. JAMA 2006;295:65–73. 9. Wu JC, Huang YH, Chau GY, et al. Risk factors for early and late recurrence in hepatitis B-related hepatocellular carcinoma. J Hepatol 2009;51:890–897. 10. Liaw YF. Antiviral therapy of chronic hepatitis B: opportunities and challenges in Asia. J Hepatol 2009;51: 403–410. 11. Yuen MF, Lai CL. Treatment of chronic hepatitis B: evolution over two decades. J Gastroenterol Hepatol 2011;26(Suppl 1):138–143. 12. Liaw YF, Sung JJ, Chow WC, et al. Lamivudine for patients with chronic hepatitis B and advanced liver disease. N Engl J Med 2004;351:1521–1531. 13. Liaw YF. Impact of hepatitis B therapy on the long-term outcome of liver disease. Liver Int 2011;31(Suppl 1): 117–121. 14. Chang TT, Liaw YF, Wu SS, et al. Long-term entecavir therapy results in the reversal of fibrosis/cirrhosis and continued histological improvement in patients with chronic hepatitis B. Hepatology 2010;52:886–893. 15. Marcellin P, Gane E, Buti M, et al. Regression of cirrhosis during treatment with tenofovir disoproxil fumarate for chronic hepatitis B: a 5-year open-label follow-up study. Lancet 2013;381:468–475. 16. Lai CL, Yuen MF. Prevention of hepatitis B virus-related hepatocellular carcinoma with antiviral therapy. Hepatology 2013;57:399–408. 17. Sherman M. Does hepatitis B treatment reduce the incidence of hepatocellular carcinoma? Hepatology 2013;58:18–20. 18. Wu CY, Chen YJ, Ho HJ, et al. Association between nucleos(t)ide analogues and risk of hepatitis B virus-

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

-,

No.

-

related hepatocellula carcinoma recurrence following liver resection. JAMA 2012;308:1906–1914. Sung JJ, Tsoi KK, Wong VW, et al. Meta-analysis: treatment of hepatitis B infection reduces risk of hepatocellular carcinoma. Aliment Pharmacol Ther 2008;28: 1067–1077. Papatheodoridis GV, Lampertico P, Manolakopoulos S, et al. Incidence of hepatocellular carcinoma in chronic hepatitis B patients receiving nucleos(t)ide therapy: a systematic review. J Hepatol 2010;53:348–356. Cho JY, Paik YH, Sohn W, et al. Patients with chronic hepatitis B treated with oral antiviral therapy retain a higher risk for HCC compared with patients with inactive stage disease. Gut 2014 Mar 10. Epub ahead of print. Niro GA, Ippolito AM, Fontana R, et al. Long-term outcome of hepatitis B virus-related chronic hepatitis under protracted nucleos(t)ide analogues. J Viral Hepatol 2013;20:502–507. Wu CY, Wu MS, Kuo KN, et al. Effective reduction of gastric cancer risk with regular use of nonsteroidal anti-inflammatory drugs in Helicobacter pylori-infected patients. J Clin Oncol 2010;28:2952–2957. Wu CY, Kuo KN, Wu MS, et al. Early Helicobacter pylori eradication decreases risk of gastric cancer in patients with peptic ulcer disease. Gastroenterology 2009;137: 1641–1648. e1
e2. Wu CY, Chan FK, Wu MS, et al. Histamine 2-receptor antagonists are an alternative to proton pump inhibitor in patients receiving clopidogrel. Gastroenterology 2010; 139:1165–1171. Cheng CL, Kao YH, Lin SJ, et al. Validation of the National Health Insurance Research Database with ischemic stroke cases in Taiwan. Pharmacoepidemiol Drug Saf 2011;20:236–242. Dehejia RH, Wahba S. Propensity score-matching methods for nonexperimental causal studies. Rev Econ Stat 2002;84:151–161. Prentice RL, Kalbfleisch JD, Peterson AV Jr, et al. The analysis of failure times in the presence of competing risks. Biometrics 1978;34:541–554. Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 1988; 16:1141–1154. Lin DY, Psaty BM, Kronmal RA. Assessing the sensitivity of regression results to unmeasured confounders in observational studies. Biometrics 1998;54:948–963. Altman DG, Andersen PK. Calculating the number needed to treat for trials where the outcome is time to an event. BMJ 1999;319:1492–1495. Papatheodoridis GV, Manolakopoulos S, Touloumi G, et al. Virological suppression does not prevent the development of hepatocellular carcinoma in HBeAgnegative chronic hepatitis B patients with cirrhosis receiving oral antiviral(s) starting with lamivudine monotherapy: results of the nationwide HEPNET. Greece cohort study. Gut 2011;60:1109–1116. Matsumoto A, Tanaka E, Rokuhara A, et al. Efficacy of lamivudine for preventing hepatocellular carcinoma in chronic hepatitis B: a multicenter retrospective study of 2795 patients. Hepatol Res 2005;32:173–184.

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944

945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003

2014

HBV Therapy and HCC Incidence

34. Eun JR, Lee HJ, Lee SH, et al. The effect of lamivudine and adefovir dipivoxil on preventing hepatocellular carcinoma in hepatitis B virus-related liver cirrhosis. Hepatology 2007;46:664A–665A. 35. Chen HP, Shieh JJ, Chang CC, et al. Metformin decreases hepatocellular carcinoma risk in a dosedependent manner: population-based and in vitro studies. Gut 2012;62:606–615. 36. El-Serag HB, Tran T, Everhart JE. Diabetes increases the risk of chronic liver disease and hepatocellular carcinoma. Gastroenterology 2004;126:460–468. 37. Davila JA, Morgan RO, Shaib Y, et al. Diabetes increases the risk of hepatocellular carcinoma in the United States: a population based case control study. Gut 2005; 54:533–539. 38. Wong VW, Chan SL, Mo F, et al. Clinical scoring system to predict hepatocellular carcinoma in chronic hepatitis B carriers. J Clin Oncol 2010;28:1660–1665. 39. Yang HI, Sherman M, Su J, et al. Nomograms for risk of hepatocellular carcinoma in patients with chronic hepatitis B virus infection. J Clin Oncol 2010;28:2437–2444. 40. Yang HI, Yuen MF, Chan HL, et al. Risk estimation for hepatocellular carcinoma in chronic hepatitis B (REACHB): development and validation of a predictive score. Lancet Oncol 2011;12:568–574.

9

1004 1005 Received September 3, 2013. Accepted March 26, 2014. 1006 Reprint requests 1007 Address requests for reprints to: Jaw-Town Lin, MD, PhD, School of Q1 1008 Medicine, Fu Jen Catholic University, No. 510, Zhongzheng Road, Xinzhuang Q2 1009 District, New Taipei City, 24205, Taiwan. e-mail: [email protected]; fax: þ886-2-23947899; or Jaw-Ching Wu, MD, PhD, Translational Research 1010 Division, Medical Research Department, Taipei Veterans General Hospital, 1011 Institute of Clinical Medicine, National Yang-Ming University, No. 155, Section 2, Linong Street, Taipei 11221, Taiwan. e-mail: [email protected]; 1012 fax: þ886-2-28745074. 1013 Q121014 Acknowledgments This study is based in part on data from the National Health Insurance 1015 Research Database provided by the Bureau of National Health Insurance, 1016 Department of Health and managed by the National Health Research Institutes. The interpretations and conclusions contained herein do not 1017 represent those of the Bureau of National Health Insurance, Department of 1018 Health or the National Health Research Institutes (National Health Insurance Research Database, Taiwan, available at: http://www.nhri.org.tw/nhird/en/ 1019 index.htm). 1020 Conflicts of interest 1021 Q3 The authors disclose no conflicts. 1022 Funding 1023 This work was supported by Taiwan’s National Health Research Institutes Q4 1024 (PH-103-PP-22), Taipei Veterans General Hospital and Department of Health (96VN-001, VN97-01, V98C1-027, DOH98-DC-1014), as well as the Center of 1025 Excellence for Cancer Research at Taipei Veterans General Hospital 1026 (MOHW103-TD-B-111-02) and National Yang-Ming University (103AC-T402, 1027 Ministry of Education, Aim for the Top University Plan). 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 Author names in bold designate shared co-first authorship.

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

CLINICAL LIVER

-

9.e1

1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121

Wu et al

Gastroenterology Vol.

-,

No.

-

Supplementary Table 1.Baseline Demographic Characteristics and Outcomes of Study Cohorts Before Propensity Score Matching Untreateda (n ¼ 47,611) Age, y, mean  SDd Sex, n (%) Female Male Follow-up years, mean  SDe Mean  SD Median (IQR) Nucleos(t)ide analogue therapy duration, y Mean  SD Median (IQR) Hepatoprotective agents, y Mean  SD Median (IQR) Concomitant drug usersf Statin NSAIDs or aspirin Metformin Major coexisting diseases Cirrhosis Liver decompensation Hypertension Diabetes COPD Acute coronary syndrome Cerebral vascular disease Renal failure Hypercholesterolemia Charlson’s score Mean  SD Median (IQR) Propensity scoreg Mean  SD Median (IQR) Events HCC occurrence Death before HCC occurrence Overall death

51.62  14.34 13,021 (27.3) 34,590 (72.7) 5.18  2.17 6.32 (3.48–7.00)

Treatedb (n ¼ 24,847) 41.95  13.70

P valuec <.001 <.001

6087 (24.5) 18760 (75.5) 3.52  2.2 3.44 (1.47–5.55)

<.001 <.001

1.42  0.73 1.41 (1.00–1.63) 1.41  1.44 0.87 (0.46–1.80) 6260 (13.1) 34,615 (72.7) 11,004 (23.1) 5179 2087 5133 4415 1829 2214 2535 676 747

(10.9) (4.4) (10.8) (9.3) (3.8) (4.7) (4.9) (1.4) (1.6)

0.74  1.11 0.34 (0.07–0.93) 1529 (6.2) 12,395 (49.9) 2017 (8.1) 3172 2121 2055 1713 491 676 506 496 209

(12.8) (8.5) (8.3) (6.9) (2.0) (2.7) (2.0) (2.0) (0.8)

<.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001

0.76  1.43 0 (0–1)

0.77  1.51 0 (0–1)

.294 <.001

0.29  0.17 0.25 (0.15–0.41)

0.44  0.17 0.45 (0.32–0.58)

<.001 <.001

1059 (4.3) 1185 (4.8) 1583 (6.4)

<.001

11,574 (24.3) 5972 (12.5) 11,865 (24.9)

<.001

COPD, chronic obstructive pulmonary disease; IQR, interquartile range; NSAID, nonsteroidal anti-inflammatory drug. a Untreated, not receiving nucleos(t)ide analogues. b Treated, receiving nucleos(t)ide analogues. c P values were compared using the c2 test and Student t test. d Age is treated as a continuous variable. e Follow-up is defined as the time of nucleos(t)ide analogue or hepatoprotective treatment. f Drug users indicate patients using drugs at least 1 day per month on average. g Age, sex, acute coronary syndrome, cerebral vascular diseases, COPD, diabetes, cirrhosis, liver decompensation, renal failure, hypertension, hypercholesterolemia, use of statins, use of NSAIDs or aspirin or coxibs, and use of metformin were included in the propensity score calculation.

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180

-

1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239

2014

HBV Therapy and HCC Incidence

9.e2

Supplementary Table 2.Interaction Analysis: Multivariable Cox Proportional Hazards Model Analysis for Risk of Hepatocellular Carcinoma Occurrence After Adding Interactions Between Therapy and Subgroup Factors

Covariates Treated vs untreated Age, per year Male sex Cirrhosis Liver decompensation Hypertension Diabetes COPD Acute coronary syndrome Cerebral vascular disease Renal failure Hypercholesterolemia Statin use NSAIDs or aspirin use Metformin use Interactions with treatment Age (per year)*treatment Male*treatment Cirrhosis*treatment Liver Decompensation*treatment Hypertension*treatment Diabetes*treatment COPD*treatment Acute coronary syndrome*treatment Cerebral vascular disease*treatment Renal failure*treatment Hypercholesterolemia*treatment Statin use*treatment NSAIDs or aspirin use*treatment Metformin use*treatment

Hazard ratio

95% CI

P value

0.08 1.05 1.48 1.60 0.84 0.81 0.99 0.71 0.67 0.41 0.89 0.83 0.54 0.60 0.92

0.06–0.11 1.05–1.06 1.34–1.55 1.46–1.76 0.74–0.95 0.72–0.92 0.87–1.13 0.57–0.88 0.55–0.83 0.30–0.55 0.70–1.13 0.55–1.24 0.46–0.64 0.57–0.64 0.82–1.03

<.001 <.001 <.001 <.001 .005 .001 .900 .002 <.001 <.001 .330 .360 <.001 <.001 .150

1.02 1.31 2.06 1.21 0.98 1.33 0.94 0.91 1.56 0.85 1.19 0.92 1.09 1.12

1.01–1.02 1.09–1.58 1.69–2.50 0.96–1.52 0.76–1.27 1.02–1.73 0.61–1.44 0.58–1.43 0.93–2.60 0.50–1.43 0.52–2.70 0.65–1.31 0.94–1.26 0.87–1.43

<.001 .004 <.001 .110 .880 .034 .760 .690 .090 .530 .690 .650 .270 .380

COPD, chronic obstructive pulmonary disease; NSAIDs, nonsteroidal anti-inflammatory drugs.

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298

9.e3

1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357

Wu et al

Gastroenterology Vol.

-,

No.

-

Supplementary Table 3.Sensitivity Analysis With Fixed Duration: Baseline Demographic Characteristics and Outcomes

1358 1359 1360 Age, y, mean  SDd 44.81  12.82 44.67  12.99 .609 1361 Sex .660 1362 Female 1117 (24.6) 1098 (24.2) 1363 Male 3428 (75.4) 3447 (75.5) 1364 e Follow-up years 1365 Mean  SD 4.26  1.37 4.63  1.10 <.001 Median (IQR) 5.00 (4.1–5.00) 5.00 (5.00–5.00) <.001 Q141366 1367 Nucleoside analogue therapy duration, y Mean  SD 1.46  0.84 1368 Median (IQR) 1.44 (1.02–1.53) 1369 Hepatoprotective agents, y 1370 Mean  SD 1.09  1.02 0.92  1.25 <.001 1371 Median (IQR) 0.71 (0.42–1.37) 0.47 (0.13–1.19) <.001 1372 Concomitant drug usersf 1373 Statin 409 (9.0) 420 (9.2) .716 1374 NSAIDs or aspirin 2869 (63.1) 2851 (62.7) .712 Metformin 529 (11.6) 548 (12.1) .559 1375 Major coexisting diseases 1376 Cirrhosis 517 (11.4) 517 (11.4) 1.000 1377 Liver decompensation 269 (5.9) 284 (6.2) .539 1378 Hypertension 381 (8.4) 352 (7.7) .281 1379 Diabetes 323 (7.1) 305 (6.7) .482 1380 COPD 107 (2.4) 94 (2.1) .392 Acute coronary syndrome 125 (2.8) 119 (2.6) .746 1381 Cerebral vascular disease 105 (2.3) 89 (2.0) .276 1382 Renal failure 82 (1.8) 79 (1.7) .874 1383 Hypercholesterolemia 53 (1.2) 45 (1.0) .477 1384 Charlson’s score 1385 Mean  SD 0.76  1.45 0.70  1.41 .031 1386 Median (IQR) 0 (0–1) 0 (0–1) 0 1387 Propensity scoreg Mean  SD 0.54  0.20 0.54  0.20 .451 1388 Median (IQR) 0.55 (0.41–0.70) 0.55 (0.41–0.70) .368 1389 Events 1390 HCC occurrence 867 (19.1) 301 (6.6) <.001 1391 Death before HCC occurrence 489 (10.8) 282 (6.2) <.001 1392 Overall death 884 (19.4) 405 (8.9) <.001 1393 1394 NOTE. Because nucleoside analogues have been reimbursed under the NHI program since October 1, 2003, we conducted 1395 sensitivity analysis by limiting index date of follow-up to between October 1, 2003, and September 30, 2005 and limiting the 1396 follow-up duration to 5 years. 1397 COPD, chronic obstructive pulmonary disease. a 1398 Untreated, not receiving nucleoside analogues. b 1399 Treated, receiving nucleoside analogues. c P values were compared using the c2 test and Student t test. 1400 d Treating age as a continuous variable. 1401 e Follow-up is defined as the period of nucleoside analogue or hepatoprotective treatment. 1402 f Drug users indicate patients using drugs for at least 1 day per month on average. 1403 g Age, sex, acute coronary syndrome, cerebral vascular diseases, COPD, diabetes, cirrhosis, liver decompensation, renal failure, hypertension, hypercholesterolemia, use of statins, use of NSAIDs, aspirin or coxibs, and use of metformin were 1404 1405 included in the propensity score calculation. 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 Untreateda (n ¼ 4545)

Treatedb (n ¼ 4545)

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

P valuec

-

1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475

2014

HBV Therapy and HCC Incidence

9.e4

Supplementary Table 4.Sensitivity Analysis With Fixed Duration: Multivariable Cox Proportional Hazards Model Analysis for Risk of Hepatocellular Carcinoma Occurrence

Treated vs untreated Age, per each incremental year Male Cirrhosis Liver decompensation Hypertension Diabetes COPD Acute coronary syndrome Cerebral vascular disease Renal failure Hypercholesterolemia Statin use NSAIDs or aspirin use Metformin use

Hazards ratioa

95% CI

P value

0.31 1.06 1.80 1.81 0.89 0.74 1.04 0.65 0.76 0.54 1.12 0.52 0.64 0.61 1.19

0.27–0.35 1.06–1.07 1.56–2.09 1.52–2.15 0.70–1.13 0.58–0.94 0.83–1.31 0.42–1.00 0.51–1.15 0.34–0.85 0.72–1.74 0.24–1.12 0.50–0.81 0.54–0.69 0.99–1.43

<.001 <.001 <.001 <.001 .340 .014 .740 .047 .190 .008 .620 .094 <.001 <.001 .060

COPD, chronic obstructive pulmonary disease; NSAID, nonsteroidal anti-inflammatory drug. a Adjusted for covariate factors, including age, sex, comorbidities, use of statins, use of NSAIDs or aspirin and use of metformin.

Supplementary Figure 1. Histograms of propensity score before and after matching. (A) Histogram of propensity scores before matching. (B) Histogram of propensity scores after matching. Untreated, CHB patients not receiving nucleos(t) ide analogues; Treated, CHB patients receiving nucleos(t)ide analogues.

Supplementary Figure 2. Sensitivity analysis with fixed duration. Because nucleos(t)ide analogues began to be covered by insurance on October 1, 2003, we conducted a sensitivity analysis by limiting the index date of follow-up to between October 1, 2003 and September 30, 2005 and limiting the follow-up duration to 5 years. Calculation and comparison of cumulative incidences in competing risk data ratios were conducted using modified Kaplan–Meier method and Gray’s method. Untreated, CHB patients not receiving nucleos(t)ide analogues; Treated, CHB patients receiving nucleos(t)ide analogues.

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534

9.e5

1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593

Wu et al

Gastroenterology Vol.

Supplementary Figure 3. Sensitivity analysis with add-on of an unmeasured confounder. This figure displays the trend of estimates for the treated hazard on covariate-adjusted Cox proportional hazards model. For example, when all the subjects in the untreated group have the add-on unmeasured confounder (the prevalence of the confounder in the untreated group is 1.0) and none of subjects in treated group has this unmeasured confounder (the prevalence of the confounder in the treated group is 0.0), the impact of antiviral therapy would be beneficial (HR ¼ 0.1, the bottom line in figure). On the contrary, when none of subjects in the untreated group has the add-on unmeasured confounder and all subjects in treated group have this confounder; then the impact of antiviral therapy would not be protective (HR ¼ 1.2, the top line in the figure).

FLA 5.2.0 DTD
YGAST59065_proof
14 May 2014
8:38 pm
ce

-,

No.

-

1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652