Treatment Patterns and Outcomes in Patients With Hepatocellular Carcinoma Stratified by Stage-Guided Treatment Categories

Purpose: The purpose of this study was to assess the survival and treatment patterns of hepatocellular carcinoma (HCC) stratified by the NCCN stage–guided treatment categories in the absence of a universally accepted staging system for HCC. Methods: Patients with HCC were identified using ICD-9 codes and inclusion in the Huntsman Cancer Institute tumor registry. Patients were stratified by the NCCN groupings around the time of diagnosis as potentially resectable or operable (RESECT), potentially transplantable (TRANSP), unresectable (UNRESECT), inoperable due to performance status (INOPER), or having metastatic (METAST) disease. Survival and treatment patterns were assessed by NCCN stage–guided treatment categories. Results: A total of 221 patients (72.9% men) with HCC were identified. At the time of diagnosis, patients were categorized as RESECT (n=28, 12.7%), TRANSP (n=33, 14.9%), UNRESECT (n=77, 34.8%), INOPER (n=40, 18.1%), and METAST (n=38, 17.2%). Staging information was not specified for 5 patients (2.3%) even after chart review. Kaplan-Meier analysis demonstrated significant differences in survival between RESECT and UNRESECT categories, and between UNRESECT and METAST categories. The median survivals in RESECT, TRANSP, UNRESECT, INOPER, and METAST categories were 594, 562, 247, 167, and 44 days, respectively. Patients considered RESECT most frequently underwent resection (61%, n=17) and patients considered TRANSP had the highest use of liver transplants (33.3%, n=11). Use of any treatment was low in the METAST (31.6%, n=12) and INOPER (60.0%, n=24) groups. Conclusions: Treatment patterns in the NCCN groupings correlated with recommended treatment strategies. Overall, the NCCN groupings have a linear relationship in overall survival. (J Natl Compr Canc Netw 2015;13:987–994) antitrypsin deficiency, tyrosinemia, and glycogen storage diseases), obesity, and diabetes.4 A defined tumor or cancer staging system is important to define survival estimates and to guide treatment options.5 HCC causes 2 pathologic changes: tumorous changes in the hepatocytes, and cirrhosis, chronic inflammation, and fibrotic changes or liver parenchyma.6 These changes have led to several different staging practices for HCC. For example, the AJCC/TNM system evaluates tumor extension and does not take into account the residual liver function,7 which may lead to inadequate staging.8 Each staging system may not apply to all patients with HCC because of differing tumor From aDepartment of Pharmacotherapy, University of Utah, Salt Lake City, Utah; bHuntsman Cancer Institute, University of Utah Health Care, Salt Lake City, Utah; cAbbVie Inc., North Chicago, Illinois; and dUniversity of Utah Program in Personalized Health Care, Salt Lake City, Utah. Submitted December 12, 2014; accepted for publication June 22, 2015. A portion of this work was supported by an unrestricted research grant by AbbVie Inc. Huntsman Cancer Institute provided partial support for all datasets within the Utah Population Database. The authors have disclosed that they have no financial interests, arrangements, affiliations, or commercial interests with the manufacturers of any products discussed in this article or their competitors. Drs. Kuo, Stenehjem, Albright, and Brixner are partially paid through the University of Utah via an unrestricted research grant from AbbVie Inc. Dr. Ray is employed by or holds a leadership position with AbbVie Inc. A portion of this study was presented at the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) 5th Asia-Pacific Conference; September 2–4, 2012; Taipei, Taiwan. Correspondence: Kuan-Ling Kuo, PhD, Department of Pharmacotherapy, University of Utah College of Pharmacy, 30 South 2000 East, Room 258, Salt Lake City, UT 84112. E-mail: Kaitlin.kuo@pharm.utah.edu Treatment Patterns and Outcomes in Patients With Hepatocellular Carcinoma Stratified by Stage-Guided Treatment Categories Kuan-Ling Kuo, PhDa; David Stenehjem, PharmDa,b; Frederick Albright, PhDa; Saurabh Ray, PhDc; and Diana Brixner, PhD, RPha,d Background Hepatocellular carcinoma (HCC) is a primary malignancy of the liver. The American Cancer Society estimates that 33,190 new cases for primary liver cancer and intrahepatic bile duct cancer were diagnosed in 2014 in the United States.1 Moreover, Tanaka et al2 reported that the incidence of HCC is expected to increase over the next 2 decades.2 Cirrhosis is the most common risk factor, which is also a consequence of hepatitis C virus (HCV) or hepatitis B virus (HBV) infections.3 Other risk factors for HCC include heavy alcohol use, Asian race, male sex with confirmed heritable metabolic diseases (Wilson disease, porphyria cutanea tarda, alpha-1


Background
Hepatocellular carcinoma (HCC) is a primary malignancy of the liver.The American Cancer Society estimates that 33,190 new cases for primary liver cancer and intrahepatic bile duct cancer were diagnosed in 2014 in the United States. 1 Moreover, Tanaka et al 2 reported that the incidence of HCC is expected to increase over the next 2 decades. 2 Cirrhosis is the most common risk factor, which is also a consequence of hepatitis C virus (HCV) or hepatitis B virus (HBV) infections. 3Other risk factors for HCC include heavy alcohol use, Asian race, male sex with confirmed heritable metabolic diseases (Wilson disease, porphyria cutanea tarda, alpha-1 morphology, and the incidence of comorbidities may vary from region to region.Additionally, the Okuda staging system is inapplicable to early HCC, because it was derived from advanced HCC cases. 9The Barcelona Clinic Liver Cancer (BCLC) system is endorsed by the European Association for the Study of the Liver (EASL) and the American Association for the Study of Liver Diseases (AASLD) 10,11 ; however, the BCLC system's acceptance is not universal. 12onsequently, geographic-based differences in HCC staging and description still exist, and validation is still an open issue for both Western and Asian-Pacific patients.
Given the absence of a universally accepted staging or descriptive standard for HCC, the 2015 version of the NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines) for Hepatobiliary Cancers classifies the severity of HCC by treatment groups as (1) potentially resectable (RESECT) or transplantable (TRANSP), ( 2) unresectable (UNRESECT), (3) inoperable based on performance status or comorbidity, with local disease only (INOPER), and (4) metastatic disease (METAST). 13ur research used these NCCN groups to assess treatment patterns and survival.

Data Source
The University of Utah Health Sciences Center enterprise data warehouse (EDW) provided the HCC research data mart for this study; an EDW derivative was based on the list of HCC cases identified in this study.Death certificates were linked via an electronic master subject computerized index link to the Utah Population Database (UPDB), which hosts state death certificates.Cases within the HCC research data mart were linked to the Huntsman Cancer Institute Tumor Registry (HCI-TR) through the same master subject index, and the HCI-TR was used to identify tumor staging and date of HCC diagnosis.The HCI-TR served to confirm 201 HCC cases.The remaining 20 study cases were confirmed via manual inspection of the case in the medical record.

Patients and Study Design
This study was a retrospective observational cohort analysis.The index date was defined as the date of HCC diagnosis in the EDW.Inclusion criteria were as follows: patients aged 18 years or older at the time of diagnosis between 1995 and 2010; at least 2 International Classification of Diseases, 9th revision (ICD-9) codes for HCC (ie, 155.0), with ICD for Oncology site and histology codes indicative of HCC in the HCI-TR, and at least 2 encounters in the EDW, with 1 encounter at least 30 days from the index date (Figure 1, available online, in this article, at JNCCN.org).Demographic characteristics (age, sex, race, plan type, region) were recorded on index date and clinical characteristics (Charlson Comorbidity Index [CCI], cirrhosis, HBV, HCV, alcohol liver, encephalopathy, nonalcoholic fatty liver disease) were identified by ICD-9 codes and assessed within 30 days of the index date.Comorbidities were assessed by the CCI in the year prior to the index date and modified by excluding the weight for any solid tumor or metastatic solid tumor diagnosis. 14Treatment patterns were evaluated according to Current Procedure Terminology (CPT) codes within the EDW for ablation, resection, transplant, embolization, and radiation.Chemotherapy use was obtained from linked pharmacy records within the EDW.Date and cause of death were obtained from the UPDB.Survival and treatments were evaluated on the date of diagnosis forward and stratified by NCCN groupings at the time of diagnosis and censored at the date of last follow-up.

Classification of Stage-Guided Treatment Categories
Stage at diagnosis was ascertained by linking the EDW to the HCI-TR, which provided AJCC TNM staging at diagnosis.All patients with stage IIIB and IIIC disease were classified as UNRESECT, and patients with stage IV (M1) disease were classified as METAST.All patients classified as stage I to IIIa were assessed by chart review to determine whether the treating physicians deemed the patient as resectable, transplantable, unresectable, or inoperable due to performance status around the time of diagnosis.To facilitate chart review, a semiautomated keyword search of the electronic notes was conducted with a specially built text search tool using keywords and Boolean constructs.Only physician notes within 45 days of diagnosis were used to determine the stageguided treatment categories.

Primary Treatment Modality
The primary treatment modality was categorized into mutually exclusive groups based on the treatment ap-proach most likely to confer a survival benefit.First, patients who received a liver transplantation at anytime during survival were identified.Then patients who received a liver resection without transplantation were selected.Of those remaining, any patient who received radiofrequency ablation, other ablation procedures (cryoablation, percutaneous alcohol injection, microwave), or transarterial chemoembolization were assigned to the liver-directed therapy group.Patients who received chemoembolization and ablation were grouped together, because these procedures are broadly are considered locoregional therapy; however, results between ablation and chemoembolization are presented separately.Lastly, patients who received only radiation therapy, systemic therapy, or no therapy were identified (Appendix 1 and Figure 1, both available online, in this article, at JNCCN.org).

Statistical Methods
Descriptive statistics were used to describe demographic characteristics, baseline clinical characteristics, and comorbidities.Chi-square test was used to evaluate the differences between the categorical variables.For time to event, median values and Kaplan-Meier survival estimates were reported.Median survival time among different NCCN groups and different treatments were calculated.
The log rank test was used to test for statistical differences in survival between groups.The variables tested included age, sex, CCI, NCCN groups, the presence or absence of cirrhosis, alcoholic cirrhosis, serologic evidence of HBV or HCV infection, encephalopathy, and nonalcoholic fatty liver disease.Multivariate Cox proportional hazards regression was used to test the relationship between various prognostic factors and survival in all patients.Variables in the COX analysis were included if they had a P value of less than .25 in the univariate analysis.All statistical tests were performed at an a priori significance level of 0.05 using Stata/SE version 10.0 (StataCorp LP, College Station, TX).The protocol for this study was approved by University of Utah Institutional Review Board and the Resource for Genetic and Epidemiologic Research at the University of Utah on November 3, 2010.

Treatment Patterns
Patients classified with RESECT disease had the highest proportion (96.4%) receiving any treatment, followed by TRANSP (81.8%),UNRESECT (80.5%),INOPER (60.0%), and METAST (31.6%) groups.The treatments received generally followed the NCCN Guidelines. 5,13The highest use of surgical resection was observed in the RESECT group and more liver transplants were performed in the TRANSP group.Reasons for not receiving a transplant in those classified as transplantable at diagnosis (n=22) included still waiting for transplant at last follow-up (n=8), size of lesions or disease progression before transplantation (n=7), poor performance status or comorbidity (n=5), and alcohol/drug abuse (n=2).The UNRESECT and INOPER groups' predominately received chemoembolization/radiofrequency ablation therapy, and the METAST group most commonly received systemic therapy and radiation therapy.Furthermore, fewer treatments were administered to the INOPER (40%) and METAST groups (68.4%).Overall, treatment patterns were distinct across the NCCN stage-guided treatment categories and aligned with NCCN treatment recommendations for these groups (Table 2).

Survival Outcomes
No statistically significant differences in the median survival were noted between the RESECT and TRANSP groups (P=.94) or between the UN-RESECT and INOPER groups (P=.79) (Figure 3).Statistically significant differences between the RESECT and METAST (P<.001) and the UNRE-SECT and METAST (P<.001) groups were observed (Figure 3 and Table 4).Patients who underwent liver transplantation had the longest median survival (30.8 months), followed by resection (21.6 months), liver-directed therapy (14.1 months), radiation therapy (9.5 months), systemic therapy (7.1 months), and no therapy (2.9 months) (Figure 4 and Table 4).Patients who only received radiation (median overall survival, 9.5 months; P=.159) or systemic treatment (median overall survival, 7.1 months; P=.003) had inferior survival outcomes compared with those who underwent surgical resection.
Because of the large survival and treatment differences in patients with metastatic disease, we excluded these patients from the Cox analysis.In the Cox analysis, UNRESECT, INOPER, younger age, and higher CCI scores (7-13) were independent predictors of reduced survival.Patients in the UNRESECT group were 2.02 times as likely to die compared with those in the RESECT group (95% CI, 1.15-3.55;P=.014).Patients in the INOPER group were 1.94 times as likely to die as those considered RESECT (95% CI, 1.04-3.62;P=.04).In addition, patients with higher  5).

Discussion
To our knowledge, this is the first study stratifying survival in HCC by the NCCN stage-guided treatment categories specified in the NCCN Guidelines.
Our study demonstrates distinct overall survival and treatments based on these categories.Our results showed no statistically significant difference in survival between the RESECT and TRANPS groups (P=.94) and the UNRESECT and INOPER (P=.79) groups, but the treatment decisions were different between these groups.Survival outcomes between the RESECT and TRANSP groups were similar, partly because only 33% of the TRANSP group received a transplant, because many were still waiting for transplant at last follow-up.Survival outcomes were improved in patients receiving a liver transplant compared with those undergoing a surgical resection alone; however, this did not meet significance, presumably because of sample size considerations.In general, the NCCN stage-guided treatment categories predicted overall survival.Similar to our studies, 15,16 resected patients or those who received liver-directed therapies had improved median overall survival (21.6 and 14.1 months, respectively).][18] Our results suggest that patients who are older had a 0.03% reduction in the risk of the death compared with those who were younger (95% CI, 0.96-0.99;P=.005); however, patients older than 65 years in our cohort were less cirrhotic (68% vs 76%) and   sample size of the study, the finding of increasing survival with increasing age should be interpreted with caution.Even though sex is not a significant factor for HCC survival in the current study (P=.407),Yang et al 21 found that the hazard ratio for overall survival of women versus men was 0.83 (95% CI, 0.77-0.88).
Farinati et al 15 stated the purpose of a staging system is to (1) provide prognostic information, (2) distinguish the treatments results, and (3) recommend the treatment in a given stage.Sirivatanauksorn and Tovikkai 6 compared 6 HCC prognostic staging systems: Okuda, 9 TNM, 8 CLIP, 22 BCLC, 10 CUPI, 23 and JIS 24 using the patients in Thailand from 2001 to 2007, and they concluded that TNM and Child-Pugh are the representative systems in patients with HCC who undergo surgical resection.Marrero et al 12 compared 7 HCC prognostic staging systems, including the above staging systems and GRETCH, 25 using the patients in University of Michigan from 2000 to 2003, and concluded that the BCLC staging system provided the best prognostic stratification for patients who have both cirrhosis and HCC, because it includes performance status, tumor extent, liver function, and treatment. 26Each system includes different characteristics and has its own advantages and disadvantages (Table 6). 12,24,26,27Based on our study, the NCCN stage-guided treatment categories provide prognostic information and distinguish treatment pathways.Therefore, NCCN stage-guided pathways can provide a reliable stratification system in HCC.
This study had some limitations.First, we cannot confirm the data accuracy in the EDW; documentation errors may be present in the electronic chart.Even though stage at diagnosis was validated through chart review, documentation errors may still exist.The major limitation of a retrospective study is the precise identification of the supposed phenotype, although this is negligible in this registry-verified study.Another problem is that, although we can assume registry cases are nearly perfect as to phenotype (stage at diagnosis), we cannot assume the same ascertainment through manual identification by the research team for the stage-guided treatment category.Second, medication and procedure data may be incomplete, because some patients may receive treatment outside the institution, which is not captured in the EDW.Also, our limited cohort size was not able to predict some of the outcomes, presumably because of low power to detect differences in smaller subgroups.In addition, a time  had a reduced incidence of HBV or HCV (35.8% vs 62.4%) compared with younger patients.Kim et al 19 reported no statistical significant difference among 5-year disease-specific survival rates between those aged 35 to 49 years, 50 to 64 years, and older than 65 years (87% vs 86% vs 85%; P=.858) after orthotopic liver transplant.Moreover, Hanazaki et al 20 found that the overall and disease-free survival rates after hepatic resections were also similar between older and younger patients (P=.95).Because of the small lag of 1 year was present in the concurrency of death certificates from the Utah Department of Health and updates to UPDB.Lastly, the clinical characteristics in our study were limited by missing/unknown data, particularly for the extent of cirrhosis and the morbidity associated with liver insufficiency.

Conclusions
Through standard review of medical notes with mitigation of variance, we demonstrate that the NCCN stage-guided treatment categories defined by the NCCN Guidelines predict overall survival and treatment use.Even though this study only used case records around the time of diagnosis for retrospective case stratification to the NCCN stage-guided groups, the categorization may be biased by the treatments received.Therefore, further validation in a prospective cohort is warranted.Additionally, because of the small sample size of the current study, future studies could be conducted across cancer centers to obtain the necessary sample size and power to fully validate the NCCN stage-guided treatment categories.

Figure 2
Figure 2 Primary treatment modality received stratified by year.

Figure 4
Figure 4 Kaplan-Meier curve among different treatment groups.

Table 2 Median Time From Diagnosis to All Treatments Received, Stratified by Stage at Diagnosis
Abbreviations: Dx, diagnosis; METAST, metastatic disease; INOPER, inoperable due to performance status; RESECT, potentially resectable or operable; TRANSP, potentially transplantable; Tx, treatment; UNRESECT, unresectable.Original Research Stage-Guided Treatment for Hepatocellular Carcinoma © JNCCN-Journal of the National Comprehensive Cancer Network | Volume 13 Number 8 | August 2015991CCI were 2.14 times as likely to die as those with a lower CCI (95% CI, 1.30-3.52;P=.003).Conversely, patients who were older had a 0.03% reduction in the risk of death compared with those who were younger (95% CI, 0.96-0.99;P=.005) (Table

Table 5 Predictors of Survival: Cox Regression Multivariate Analysis (Excluding Patients With Metastatic Disease)
BCLC, Barcelona Clinic Liver Cancer; CLIP, Cancer of the Liver Italian Program; CUPI, Chinese University Prognostic Index; GRETCH, Groupe d'Etude et de Traitement du Carcinome Hepatocellulaire; JIS, Japan Integrated Staging; MELD, Model For End-Stage Liver Disease; NA, not available.
© JNCCN-Journal of the National Comprehensive Cancer Network | Volume 13 Number 8 | August 2015