| |
HCC Mismanaged in Medicare Population
|
| |
| |
Treatment and outcomes of treating of hepatocellular carcinoma among Medicare recipients in the United States: A population-based study
Journal of Hepatology
Jan 2006
Hashem B. El-Seragab, Abby B. Siegeld, Jessica A. Davilaa, Yasser H. Shaibab, Mikele Cayton-Woodya, Russell McBridee, Katherine A. McGlynnc
a Section of Health Services Research, Houston Veterans Affairs Medical Center, Baylor College of Medicine, Houston, TX, USA
b Section of Gastroenterology, Houston Veterans Affairs Medical Center, Baylor College of Medicine, Houston, TX, USA
c Division of Cancer Epidemiology and Genetics, NCI/DHHS, Columbia University College of Physicians and Surgeons, New York, NY, USA
d Department of Medicine and the Herbert Irving Comprehensive Cancer Center, Columbia University College of Physicians and Surgeons, New York, NY, USA
e Mailman School of Public Health, Columbia University College of Physicians and Surgeons, New York, NY, USA
Note From Jules Levin: The study from El-Seraga finds the HCC patient study population was mismanaged. Why? The Editorial, written by Spanish researchers, appears not to understand why. Perhaps the journal should have asked an American to write the Editorial. I think the mismanagement is a result of several factors: (1) long before these patients develop HCC, they were likely not receiving good overall care/treatment for their health, co-mordidities, and for their liver disease. It is well known that in the USA liver disease is often not properly managed, in part because general health practictioners are not well schooled in liver disease are managing patients improperly. Clearly, the similar mismanagement occurs in other marginalized & disenfranchised patient populations such as HIV coinfection, Blacks & Latinos, & IVDUs. (2) once these patients, many of whom are poor & ignorant about liver disease & comorbidities, progress to later stage liver disease they do not know how to handle their diseases and their heathcare providers suffer from both ignorance and neglect regarding their patient's interests.
EDITORIAL Comments: "....HCC is an emerging health problem that has not been adequately dealt with in several countries .... the economic burden of HCC management is not negligible...hospitalisation of HCC patients in the US during the year 2000 may have had a cost of $509 million, a figure that is compatible with the estimate made by the American Gastroenterological Association, that calculated all the costs associated with HCC in 1998 to be $998 million...The results (of this study) should represent a concern for the authority responsible of health care delivery to HCC patients within Medicare. There is significant heterogeneity in the type of treatment offered and in some instances the therapy applied does not adhere to the most recent state of the art recommendations )from EASL and AASLD).....the problem has unequivocally emerged in the US and the study by El-Serag et al. shows that the healthcare delivery to HCC patients within Medicare is not optimal...."
Author Discussion
This is the first population-based study of the extent and determinants of HCC therapy in the United States and the outcomes of these therapies. Three main findings indicate potentially significant inappropriate management of HCC during the years 1992-1999. First, the great majority of patients with HCC did not receive potentially curative therapy. More importantly, only a third of patients with favorable tumor features who were most likely to benefit received such therapy. Second, potentially inappropriate use of curative therapy, mostly resection, was observed in approximately a fifth of patients with unfavorable features such as lesions >10cm [3]. Lastly, there were remarkable geographic variations indicative of wide practice variations in the extent and type of curative, as well as palliative, therapies.
Given the lack of population-based studies, the acceptable proportion of HCC patients in whom potentially curative or palliative therapy should be applied is not known. Estimates from non population-based large referral centers such as the Barcelona Clinic Liver Center, indicate that 28% of 2114 consecutive patients HCC presenting between 1987 and 2002 were treated with potentially curative therapy (resection 6%, transplantation 9%, ablation 13%) [10]. Data from the Cancer of the Liver Italian Program (CLIP) on 650 patients diagnosed between 1990 and 1997 indicate that 41% received liver resection or local ablation, and 16% received TACE/TAE [11]. In the CLIP data, about 31% of patients 70 years and older received potentially curative therapy. Neither study was a true population-based study. Nevertheless, our results, with only 13% receiving potentially curative therapy suggest marked underutilization of such therapy.
There are several possible explanations for this apparent underutilization of HCC therapy. The severity of liver disease, comorbid illness, and functional status are important determinants of treatment and prognosis of patients with HCC. In this study, we relied on diagnostic codes of ascites, encephalopathy, and bleeding esophageal varices to define decompensated liver disease because laboratory testing and imaging studies were not available. We also estimated disease comorbidity. However, it is possible that there was residual unmeasured comorbidity that would explain some of the observed underutilization of therapy.
Our findings indicate remarkable geographic variations in the extent and type of curative and palliative therapies independent of demographic, clinical, and tumor features. The lack of a uniformly accepted standardized staging system for HCC could have contributed to these findings. Moreover, except for the Barcelona Clinic Liver Cancer staging system, the other staging systems are not directly coupled to treatment modalities. Lastly, healthcare providers' experience with diagnosis and treatment of HCC was unlikely to be great, especially in the earlier years of the study period due to the relative infrequency of this cancer at that time.
Our data also show that the most effective treatment option for HCC is liver transplantation. Transplantation offered the best chance for long-term survival, and when efforts to adjust for propensity score were included, the mortality risk was two-thirds less as compared with surgical resection. Compared with the Mazzaferro data, which yielded a 4-year survival of 75%, our 3-year survival with transplantation was only about 40% [12]. A much higher percentage of our patients had Child's C cirrhosis, and our patients were also significantly older than those in Mazzaferro's study (median age was 52). It is also unclear how many patients of ours received transplants outside of established criteria. Previous studies support our finding that ablation seems to be more effective than TACE [13,14]. However, our data suggest in a population-based sample that neither is an effective long-term curative strategy. Survival with ablation was not different from resection in the first year but dropped off quickly after this.
These findings have to be interpreted within the potential limitations of our study. This study included only Medicare-enrolled patients, so most patients were 65 years and older. Thus, generalization to younger patients is limited. However, data from SEER registries indicate that 55.4% of HCC patients are 65 years and older. The use of diagnostic and procedure codes to identify therapy may also carry some variability depending on the facility and providers. Further, the study period preceded the introduction of the MELD scores in 2000, and the likely wider use of RFA. Estimation of average treatment effects in observational studies requires adjustment in pre-treatment variables. Rosenbaum and Rubin proposed an alternative method for adjusting for pre-treatment variables based on the propensity score, which is the conditional probability of receiving treatment given pre-treatment variables [9]. Therefore, we used propensity scores to mimic randomization; however in the absence of true randomization, baseline differences between the groups could still account for the observed differences in survival.
In summary, the findings of this population-based study of predominantly 65 years and older patients with HCC indicate wide practice variations in management. Of particular concern is the apparent underutilization of potentially curative therapy in patients with favorable tumor features. The barriers to implementing appropriate treatment should be identified and strategies for increasing the utilization of these therapies should be developed. We found that transplantation offers the best chance for long-term survival. Resection offers the next best survival, followed by ablation, and finally TACE. Unlike previous studies, which have suggested that ablation is a potentially curative treatment, here it yielded a 3-year survival of only about 10%.
EDITORIAL
Journal of Hepatology Jan 2006
Maria Varela, MD, Jordi Bruix, MD
BCLC Group, Liver Unit, Hospital Clinic, University of Barcelona, IDIBAPS, Villarroel 170, 08036 Barcelona, Catalonia, Spain
As is commonly known, HCC appears in most cases within a chronically diseased liver, the most frequent etiologic agents being hepatitis B and C viruses and excessive alcohol intake. Interestingly, modern cohort studies have shown that HCC is now the leading cause of death in patients with cirrhosis [3-5]. Hence, all hepatologists and healthcare providers involved in the management of patients with liver disease are now aware of the increased risk of cancer of their patients. Simultaneously, major effort and importance is placed in developing guidelines and protocols to deliver optimal healthcare to patients with either suspected or proven HCC. Years ago, the European Association for the Study of Liver Disease (EASL) organised the Barcelona Monothematic Conference, where a panel of experts produced the first Western document to guide the diagnosis and treatment of patients with HCC [6]. The American Association for the Study of Liver Diseases (AASLD) has recently published the Practice Guidelines for the Management of HCC [7] and in the following months a new document prepared by a panel of experts of EASL, AASLD and the Japan Society of Hepatology that met at the second Barcelona meeting on HCC will further homogenise the management and research needs on a more worldwide level.
These comments are relevant to frame the value and message of the study by the group of El-Serag published in this issue of the Journal [8]. This team has played a major role in the definition of the importance of HCC in the United States and the specific epidemiological profile in their country. Their data have unequivocally depicted the increase in the incidence of HCC (which has almost doubled in the last 30 years) [9] and through the analysis of large databases they have emphasised the role of viral infection, alcoholism, diabetes and HIV [10-12]. The new study further digs into the databases and tries to assess to what extent the management of the HCC patients is adequate and/or homogeneous and whether the outcome of the patients fits into the expected figures. The results should represent a concern for the authority responsible of health care delivery to HCC patients within Medicare. There is significant heterogeneity in the type of treatment offered and in some instances the therapy applied does not adhere to the most recent state of the art recommendations. In addition, the survival registered in the database does not reproduce the findings of modern cohort studies in which treatment allocation is decided following a proper algorithm. El-Serag et al. show that only 11% of the potential candidates for transplantation were actually transplanted; that just 13% patients for surgical resection received such therapy and, finally, that only 14% of those apparently fit for local ablation were treated. By contrast, 19% of patients with HCC lesions >10cm and 5% of patients with metastatic disease underwent therapies with curative intent, while in most referral Units they would have been dismissed because of the well-known poor outcome [8]. As a whole, it appears that some individuals who could have benefited from therapy were not treated and some who were treated would have been better served if they had been left untreated.
Obviously, all this rough interpretation should be tempered because of the database limitations. The most important of them is the fact that the database represents a non-random segment of the population, whose healthcare is provided by Medicare. Furthermore, the database allows to extract a general view of the situation regarding HCC management, but the data collected to construct the database does not allow to perform an in depth analysis to clarify why treatments were not offered in some instances or why they were apparently indicated in a controversial strategy. The retrospective analysis of the database cannot ensure a proper diagnosis and staging of the patients and obviously these critical points in clinical decision making were largely heterogeneous among centres. Furthermore, significant co-morbidity cannot be accurately established and this may be the most important reason to avoid therapy. The authors have used an index to estimate presence of co-morbidities, but even with this approach, the reason why treatment was offered or denied is impossible to be ascertained. Intriguingly enough, the authors show that there are also major differences according to the location of the patients. This means that the treatment strategy is not the same all over the US, but the difference may also be the result of some specific cultural or economic profile of the population attended by Medicare in different states.
It could be argued that population based studies assessing incidence, treatment allocation and survival are not available. Hence, the study by El-Serag et al. may reflect real life while cohort studies raised in referral Units merely report the cases in best conditions that are evaluated in these specific Centres. While this could be partially the case in the provision of treatment and for overall survival, the argument cannot be used to explain the outcome after therapies. Current data applying adequate criteria indicate that survival after resection, transplant or percutaneous ablation should definitely exceed 50% at 3 years [7] and here we are given figures that are far below these limits. Median survival after transplantation is approximately less than 4 years, and does not reach 3 years after surgical resection. Similarly, median survival after percutaneous ablation and transarterial chemoembolization does not reach 2 and 1 year, respectively.
Do these comments suggest that the study by El-Serag et al. has no value? Not at all! It is clear that HCC is an emerging health problem that has not been adequately dealt with in several countries. The present study indicates that this is also true at least in some specific populations within the US. This critical information should be the trigger to set up the strategy to further assess the gaps between state of the art knowledge and its application in real life. Both EASL and AASLD have made the effort to offer health care providers with guidelines to deliver optimal care [6,7]. Furthermore, the management and research in the US has been the focus of specific conferences organised by AASLD and NIH, that have prompted that liver cancer is now one of the targets of a major research funding within the NIH action plan of the liver disease section. This research activity both in the experimental laboratory and in the clinical setting should be instrumental to face the HCC 'epidemic' with proper cost-efficiency. In that sense, it has to be stressed that the economic burden of HCC management is not negligible. A recent analysis by Kim et al. [13] has shown that the clinical activity related to HCC management has markedly increased in the past 20 years and that this has a direct translation in costs. According to these authors, the hospitalisation of HCC patients in the US during the year 2000 may have had a cost of $509 million, a figure that is compatible with the estimate made by the American Gastroenterological Association, that calculated all the costs associated with HCC in 1998 to be $998 million [13].
In summary, while great improvements have been achieved in the awareness of the relevance of HCC and its increasing incidence, there is still a major need to improve all the aspects related to its diagnosis and management.
Back To the Published Study
Background
The incidence of hepatocellular carcinoma (HCC) has doubled over the last 20 years, with a substantial proportion of this increase attributed to hepatitis C [1]. The overall prognosis of patients with HCC in the US is poor, especially for patients who do not receive specific therapy [2]. Potentially curative therapy for HCC includes surgical resection, liver transplantation, and possibly local ablation with alcohol and radiofrequency [3]. These therapies have been shown in uncontrolled series to be associated with longer survival than expected without therapy, particularly among patients with smaller tumor size, fewer lesions, and less severe liver disease [4]. In addition, some forms of palliative therapy such as trans-arterial chemoembolization (TACE) have also been shown in a recent meta-analysis of randomized controlled trials to be associated with longer survival in patients with relatively preserved liver function [5]. However, most of these studies evaluated a small number of patients, and focused on selected patient populations.
The outcomes of HCC in the US population are unclear due to the lack of population-based data on HCC therapy. These outcomes depend on the effectiveness of therapy but also on the extent of using these therapies for HCC. Establishing estimates of the extent of diffusion of therapies is important in determining the effectiveness of treatment and in identifying gaps in the equity of care. The registries of the Surveillance, Epidemiology, and End Results (SEER) Program collect population-based cancer incidence and survival data from different sites across the country [6]. The SEER-Medicare database merges SEER and Medicare, and contains demographic, clinical and medical claims data including treatment on cancer patients mostly over age 65 at diagnosis [7]. It has been extensively used to examine the outcomes of therapy for several cancers but not liver cancer. Using SEER-Medicare, we examined the extent and potential determinants of receiving (and type) of treatment, and the effects of receiving different modalities on survival of patients with HCC.
ABSTRACT
Background/Aims: There are several treatment alternatives available for patients diagnosed with hepatocellular carcinoma (HCC). Yet, neither the extent to which potentially curative or palliative therapy is used to treat HCC, nor the determinants of using such therapies are known. Further, it is unclear how effective different modalities are for treating HCC.
Methods: We used the linked SEER-Medicare dataset to identify patients diagnosed with HCC between 1992 and 1999.We identified 2963 patients with continuous Medicare enrollment who were not enrolled in a Medicare-HMO. HCC treatments were categorized as potentially curative therapy (resection, transplant, local ablation), or palliative (trans-arterial chemoembolization (TACE), chemotherapy), and no therapy. Demographic (age, sex, race, geographic region), clinical (comorbidity, risk factors and severity of liver disease) and tumor factors (tumor size, extent of disease) were examined as potential determinants of therapy, as well as survival in univariate and multivariable analyses. Survival curves were also generated and compared among the different treatment modalities.
Results: The median age at diagnosis was 74 years (range: 32-105), and most patients (91%) were older than 65 years. Approximately 68% were White, 10% Black, 4% Hispanic, 8% Asian, and 9% were of other race. Thirteen percent of the patients received potentially curative therapy (transplant 0.9%, resection 8.2%, local ablation 4.1%), 4% received TACE, 57% received other palliative therapy, and 26% received no specific therapy. Only 34% of 513 patients with single lesions, and 34% of 143 patients with lesions <3.0cm received potentially curative therapy. However, 19.2% of patients with unfavorable tumor features (lesion >10.0cm) received such therapy. Among patients who received potentially curative therapy (n=392), resection was the most common procedure (n=243, 62%) followed by local ablation (n=122, 31%) and finally transplantation (n=27, 7%). In regression analyses, geographic variations in the extent and type of curative therapy persisted after adjusting for demographic, clinical, and tumor features. Median overall survival was 104 days following HCC diagnosis with the longest survival in the transplant group (852 days) and the shortest survival in the group with no treatment (58 days). In the survival analysis, transplantation led to the longest survival, followed by resection. Neither ablation nor TACE yielded prolonged survival (3 year survival was less than 10%).
Conclusions: In this predominantly 65 years and older Medicare population, there are marked geographic variations in the management of HCC that seem to be at least as important as clinical and tumor-related features in determining the extent and type of HCC therapy. There is underutilization of potentially curative therapy, even among those with favorable tumor features.
Methods
Data source
The SEER-Medicare dataset contains Medicare claims data dating back to 1991 for all Medicare-enrolled patients identified by SEER registries between 1992 and 1999.SEER collects population-based cancer incidence and survival data on incident cancer cases from 11 population-based cancer registries that account for approximately 14% of the population in the US. Medicare is the primary health insurer for approximately 97% of individuals age 65 years and older in the US. Persons less than 65 years of age are eligible for Medicare benefits if they are disabled or have end stage renal disease.
Study population
All Medicare-enrolled patients with a diagnosis of HCC in SEER registries between 1992 and 1999 were eligible for inclusion. Diagnostic confirmation of HCC was defined as having positive histology, cytology, laboratory test/marker study, direct visualization of tumor or a positive radiology test. We excluded patients diagnosed with stomach, colon, rectum, lung, pancreas, or breast cancers within the 5 years prior to the date of HCC diagnosis to avoid metastatic liver cancers.
To study patients with equal exposure to risk factor information, we selected only those with continuous enrollment in Medicare for at least one year prior to HCC diagnosis. We also excluded patients enrolled in a health maintenance organization (HMO) during this time period because Medicare HMO plans have not been required to submit individual claims to CMS for specific services received by patients enrolled in Medicare.
HCC treatment
Transplantation had the highest precedence, followed by resection, ablation, and TACE. Patients with none of these procedures were grouped into a separate category.
Risk factors for liver disease
HBV, HCV, diabetes, and alcoholic liver disease were identified from inpatient and outpatient files from 1 year preceding and for 2 years succeeding the date of HCC diagnosis or until death.
Disease comorbidity
To estimate the severity of liver disease, we identified the following conditions during the 1-year prior to HCC diagnosis: encephalopathy, ascites, esophageal varices, and hepatorenal syndrome.
We also identified patients with Child C cirrhosis based on an algorithm derived from a dataset of 159 patients with newly diagnosed HCC in whom the Child score was calculated based on medical record review. A logistic regression model predicting 'Child C' was performed with ascites, encephalopathy, alcoholism, HCV, HBV, cirrhosis, and CT of the abdomen as predictor variables, and a C-statistic of 0.75 (indicates the predictive ability of the model) was reached. Parameter estimates obtained for the seven predictor variables in the fitted model was attached to each observation of the current SEER-Medicare HCC cohort (n=2963) and a logistic regression model was used to calculate the Child Class score variable in this cohort. The score was converted to a categorical variable with probability ≥0.28 indicating Child C; this cutoff is associated with a negative predictive value of 82%.
In addition, we constructed a general disease comorbidity index based on outpatient and inpatient diagnoses recorded within one year prior to the diagnosis of HCC. We followed the methods previously described and validated by Klaubunde using the linked SEER-Medicare dataset [8].
Results
There were 2963 patients with HCC diagnosed between 1992 and 1999 who fulfilled our inclusion and exclusion criteria. The method of diagnosis was histology in 63%, cytology in 20%, abnormal lab test in 2%, direct visualization in 1% or radiology tests in 14%. Approximately, two-thirds of patients (62.5%) had codes indicative of cirrhosis or its complications. The median age of patients was 74 (range: 32-105), and most patients (91%) were older than 65 years. Men comprised 68% of these patients. The racial composition was 68% White, 10% Black, 4% Hispanic, 8% Asian, and 9% other race. Of those, 27 (0.9%) received liver transplantation, 243 (8.2%) surgical resection, 122 (4.1%) local ablation, 131 (4.4%) TACE, 57% other palliative therapy (12% systemic chemotherapy and 45% radiotherapy), and 26% received no specific HCC therapy.
There were significant differences in HCC treatment according to tumor-related and clinical factors (Tables 1 and 2). Approximately, 28% of the patients with more than one lesion received no therapy, compared to 15% of patients with a single lesion. Patients infected with HBV were more likely to receive curative therapy than patients with other risk factors, followed by those with HCV. During the latter half of the study period, there were small increases in the proportions of patients receiving potentially curative therapy (15.0%) or TACE (5.1%). The proportions of patients treated with curative therapy ranged from 8.7% in Connecticut to 20.6% in Hawaii. Approximately, a quarter of Asians with HCC received potentially curative therapy, which was significantly greater than the proportions in other racial groups (range: 11.8-14.8%).Table 1.
However, in the full multivariable logistic regression model, Asian race was not an independent determinant of receiving potentially curative therapy (Table 3). Similarly, there were no significant temporal changes in the receipt of curative therapy when adjusted for the variables mentioned above. The presence of multiple lesions (65%), metastatic disease (-84%), tumor size >5.0cm (-30%), and diabetes (-27%) were significant negative determinants of receiving potentially curative therapy, whereas HCV (+20%) and HBV (+218%) were significant independent positive determinants of obtaining such therapy. Lastly, significant geographic variations persisted in the fully adjusted model.
Among 392 patients who received potentially curative therapy, the majority received surgical resection (62%), followed by local ablation (31%), and liver transplant (7%). There were no significant differences in sex, or race. However, there were significant geographic differences related to the type of curative therapy. All three types of curative therapy were performed more frequently in patients from the Los Angeles registry. In the full multivariable logistic regression model, the more recent time period (-48%), multiple HCC lesions (-47%), metastatic disease (-69%), and severe underlying liver disease (-83%) were significant independent negative determinants, whereas HCC size >5.0cm (+214%) was a positive determinant of surgical resection (Table 4).
We identified 96 patients who hypothetically were ideal candidates for transplantation: patients younger than 70 with one mass <5cm HCC or <3 tumors. Of these, 11 (11.5%) had transplant, 15 (15.6%) had surgical resection, and 13 (13.5%) had local ablation. We also identified 124 patients who might have been good candidates for surgical resection (size <10cm and absence of codes for cirrhosis or hepatic decompensation), and of those 12.9% received surgical resection. Lastly, of 56 patients with tumor size <3cm and single or multiple lesions (but not metastatic or unknown) who might have been good candidates for local ablation, only 14.3% received such therapy. Interestingly, a relatively large proportion (19.3%) of patients with unfavorable tumor features for potentially curative therapy (HCC lesion(s) >10.0cm) received such therapy; most of these patients (81.6%) received surgical resection. In addition, 4.9% of those recorded as having metastatic disease received a form of potentially curative therapy.
There were significant differences in survival among the treatment groups (Table 5 and Fig. 1). The median overall survival was 104 days following HCC diagnosis with the longest survival in the transplant group (852 days) and the shortest survival in the group with no treatment (58 days). The risk of mortality was reduced by 32% in patients who received transplant as compared to resection. In a Cox proportional hazard model that adjusted for propensity score, the mortality risk was reduced even further (by approximately 65%) with transplant compared to surgical resection.
In the Kaplan Meier analysis, the cumulative survival of patients who received surgical resection was significantly higher than ablation (P=0.005). The unadjusted Cox model showed a 35% mortality reduction with resection compared to ablation. However, in a Kaplan Meier analysis of 69 patients in each group matched on propensity score to receive surgical resection, these differences in survival fell short of statistical significance (P: 0.08). Similarly, in the adjusted full model, surgical resection was no longer a significant predictor of lowered mortality (Table 6). Most of the adjustment resulted from including the propensity score variable. Further, in a conditional logistic regression model that examined only 64 patients in each group matched on propensity score, the unadjusted hazard ratio with surgical resection was also not significant (0.90 (0.53, 1.52)) (data not shown).
The cumulative survival in patients who received either surgical resection or TACE in the entire group, as well as 68 patients from each group matched on propensity score were then compared. In the unadjusted Cox proportional model, there was a 32% lower mortality risk with surgical resection compared with TACE. The survival benefit with resection persisted but was attenuated in a model that adjusted for the propensity score, as well as in the full model (data not shown).
Fig. 1 shows a higher cumulative survival in patients who received ablation compared to those who received TACE (P=0.008). These differences persisted (P=0.012) in analyses of 65 patients in each group matched on propensity score. Ablation was associated with approximately 30% reduction in mortality risk in the unadjusted as well as in the fully adjusted Cox proportional model (Table 7). In a conditional Cox proportional hazard analysis limited to 65 patients in each group matched on propensity score, the hazard ratio was 0.55 (95% CI: 0.32, 0.94).
Apart from the type of therapy, other significant predictors of increased mortality risk were the presence of distant disease and more comorbidities. On the other hand, gender, race, age, and year of HCC diagnosis were not significant predictors of mortality.
|
|
| |
| |
|
|
|
