Background

National [17] and international [8, 9] recommendations for hepatitis B virus (HBV) screening before chemotherapy emphasize the need to identify patients with HBV infection so that antiviral prophylaxis can be initiated to prevent reactivation of HBV infection. The pooled (range) incidence of HBV reactivation, HBV-related hepatitis, HBV-related liver failure, and HBV-related death among cancer patients receiving chemotherapy, who had not received antiviral prophylaxis has been reported to be 37% (24-88%), 33% (24-88%), 13% (5-33%) and 7% (0-63%), respectively [10]. The 7 national recommendations, however, differ regarding which patients should be screened [17] (Table 1). Unfortunately, no population-based studies have been conducted in the US to inform an evidence-based HBV screening policy.

Table 1 National recommendations

Although the rates of HBV screening before immunosuppressive therapy in the US are unknown because of the lack of large-scale studies, rates have been estimated through physician surveys, which have shown rates ranging from 38-80% [1113]. However, these studies were limited by the potential for recall bias and low survey response rates (5-63%). Furthermore, screening practices reported in these studies may not reflect actual screening practices.

We previously found that only 17% of patients treated at a US cancer center from 2004 through 2007 were screened for HBV infection before chemotherapy [14]. The purpose of this study was to update our previous study by determining HBV screening rates at the same cancer center from 2004 through 2011 and to examine the possible influence of national recommendations published between 2004 and 2010 on HBV screening rates over time.

Methods

Data sources

We conducted a retrospective cohort study of adults with newly diagnosed cancer who registered at The University of Texas MD Anderson Cancer Center between January 1, 2004, and April 30, 2011, and received chemotherapy. This study was approved by the MD Anderson Institutional Review Board. We merged patient data from 4 institutional sources:

  1. 1.

    Tumor Registry: patient demographics, including birthplace, and cancer type (hematologic malignancies vs. solid tumors); primary liver cancer was excluded because of the etiologic relationship between HBV and hepatocellular carcinoma. At MD Anderson, patient’s race/ethnicity can be ascertained based on self-reporting, reporting by the referring clinic, or assignment by administrative staff. We categorized race/ethnicity as White, Black, Hispanic, Asian, or Other. Birthplace in a region of moderate to high prevalence of HBV infection was considered a risk factor for HBV infection [3].

  2. 2.

    Pharmacy Informatics: chemotherapy drugs and dates administered. Chemotherapy was classified according to American Cancer Society classification [15]. We excluded oral chemotherapy because we could not validate medication dispensing dates. We excluded patients in therapeutic clinical trials since some clinical trials excluded patients with liver disease or hepatitis and screening for HBV was often dictated by the protocol and not reflective of the investigators’ decision.

  3. 3.

    Patient Accounts: ICD-9 codes corresponding to risk factors for HBV infection (see Table 2) anytime before the end of the screening period (defined below).

  4. 4.

    Laboratory Informatics: test dates and results for hepatitis B surface antigen (HBsAg), antibody to hepatitis B core antibody (anti-HBc), alanine aminotransferase, total bilirubin, and HBV DNA.

Table 2 Characteristics of the study population by screening status

HBV screening and infection

Screening was defined as having both HBsAg and anti-HBc tests ordered in the period from 2 months before the first administration of chemotherapy until the second administration of chemotherapy. MD Anderson has no official policy recommending prechemotherapy HBV screening. Positive findings on both HBsAg and anti-HBc tests were considered to indicate chronic HBV infection. Negative HBsAg test but positive anti-HBc test were considered to indicate occult HBV infection or convalescence after previous infection. Unfortunately, antibody to hepatitis B surface antigen (anti-HBs) test, which is positive in convalescence and negative in occult disease, was ordered in only 1% of patients.

Three time periods

We used dates of the publication of national HBV recommendations to create 3 time intervals and categorized patients in these intervals according to date of first chemotherapy administration. We used 90 days after publication of recommendations as cut-off dates to allow adequate time for dissemination and potential change in practice patterns.

  • Period 1: January 1, 2004, through December 18, 2008 (includes publication of Food and Drug Administration [FDA] letter [1] and 2007 American Association for the Study of Liver Diseases [AASLD] [2] recommendation).

  • Period 2: December 19, 2008, through September 30, 2010 (includes publication of Centers for Disease Control and Prevention [CDC], [3] 2009 AASLD, [4] National Comprehensive Cancer Network [NCCN], [5] Institute of Medicine [IOM] [6] recommendations, and American Society of Clinical Oncology [ASCO] provisional clinical opinion [PCO] [7]).

  • Period 3: October 1, 2010, through April 30, 2011 (after publication of above recommendations).

Statistical methods

We calculated screening prevalence for each time period and tested for an increase in screening across the 3 periods using Cochran-Armitage trend tests. We compared characteristics of screened and unscreened patients using Pearson’s chi-square tests for categorical variables and Student’s t-test for continuous variables. We calculated screening rate per quarter and determined the rate of change of screening prevalence per quarter by cancer type and time period using regression analysis. Our main outcome variable was screening using HBsAg and anti-HBc tests. Independent variables included age, gender, race/ethnicity, US residency, having an HBV risk factor, cancer type, rituximab therapy, and date of first chemotherapy administration. We used 2 logistic regression models to identify predictors of screening, one for patients with solid tumors and one for patients with hematologic malignancies. We used backward elimination to select final models with a criterion of P > 0.05 for exclusion. Hosmer and Lemeshow goodness-of-fit tests were used to evaluate model fit. We determined the proportion of positive test results among screened patients and compared the rates of either a positive HBsAg test or a positive anti-HBc test result across the 3 time periods using Pearson’s chi-square test. We used SAS software, version 9.2 (SAS Institute, Cary, North Carolina), for statistical analyses.

Results

During the study period, 141,877 new patients were registered at MD Anderson (Figure 1), of whom 18,688 (13.2%) received chemotherapy at MD Anderson. Overall, 3020 (16.2%) of the patients who received chemotherapy were screened for HBV infection around the onset of chemotherapy.

Figure 1
figure 1

Study patient population. Flow diagram for study patients showing the exclusion of patients who did not have chemotherapy, had investigational chemotherapy, or non-parenteral routes of chemotherapy. Patients with primary liver cancer were also excluded.

The prevalence of HBV screening was approximately 4% (581/15,031) among patients with solid tumors and nearly 67% (2439/3657) among patients with hematologic malignancies. Nearly 29% (5391) of all patients had a risk factor for HBV infection, and less than 19% of these patients (1016) were screened. Over 10% (1977) of all patients received rituximab, and nearly 69% of these patients (1360) were screened. About 15% of the Asian patients and 12% of the Black patients were screened compared to nearly 17% of the White patients (Table 2).

The prevalence of HBV screening increased slightly across the 3 time periods, from 14.8% in period 1 to 18.2% in period 2 and 19.9% in period 3 (P < 0.001) (Table 3). For patients with known risk factors for HBV infection, screening prevalence increased over the 3 periods. For patients who received rituximab, screening prevalence increased between periods 1 and 2 and then decreased slightly in period 3. For Asian patients, screening prevalence did not change significantly over the 3 periods; for Black patients, screening prevalence increased over the 3 periods (Table 3).

Table 3 Rates of HBV screening a by screening period

Screening was almost always performed with both HBsAg and anti-HBc. Rates of use of the HBsAg test alone were 0.8% in period 1, 0.4% in period 2, and 0.9% in period 3 (P = 0.06). Among the 3020 screened patients, 252 (8.3%) had a positive result for either HBsAg or anti-HBc test. Specifically, 31 (1.0%) had positive results on both HBsAg and anti-HBc tests, 218 (7.2%) had a negative HBsAg test and a positive anti-HBc test, and 3 (0.1%) had a positive HBsAg but negative anti-HBc. Assuming that unscreened patients had negative tests, the proportions of patients with a positive result on either HBsAg or anti-HBc testing among all patients who received chemotherapy in periods 1, 2, and 3 were 1.4% (169/11,833), 1.5% (84/5703), and 1.7% (19/1152), respectively (P < 0.0001).

Solid tumors

Among patients with solid tumors, screening rates for periods 1, 2, and 3, respectively, were as follows: breast: 2.1%, 2.2%, 5.8% (P = 0.01); lung: 1.1%, 2.9%, 2.9% (P = 0.009); colon: 3.5%, 4.6%, 3.4% (P = 0.30); and prostate: 2.8%, 2.9%, 2.8% (P = 0.37). The odds of HBV screening were increased by 30% and 70% for patients who had chemotherapy in periods 2 and 3, respectively, compared to period 1 (Table 4). Other significant predictors of higher rate of HBV screening were younger age, male gender, US residence, having at least 1 HBV risk factor, and planned rituximab therapy. HBV screening was performed in 64.4% of patients with and in 3.2% without rituximab in their chemotherapy regimen.

Table 4 Predictors of HBV screening a by cancer type

Hematologic malignancies

Among patients with hematologic malignancies, the screening rate increased during period 1 by 1% per quarter and then stabilized for periods 2 and 3 (Figure 2). This pattern was seen in lymphoma patients (63.5%, 81.3%, 81.3% for periods 1, 2, and 3, respectively; P ≤ 0.001) and acute leukemia patients (75.2%, 88.9%, 89.1% for periods 1, 2, and 3, respectively; P ≤ 0.001). The odds of screening were nearly twice as high for patients who had chemotherapy in period 2 as for patients who had chemotherapy in period 1 (Table 4). No incremental effect was observed after publication of national recommendations. Other significant predictors of screening were younger age, having at least 1 HBV risk factor, and planned rituximab therapy. HBV screening was performed in 69.2% of patients with and 64.2% of patients without rituximab in their chemotherapy regimen. Black race was associated with a lower screening rate.

Figure 2
figure 2

Trends in HBV Screening at MD Anderson Cancer, 2004–2011, in relation to publication of recommendations. HBV screening prevalence is shown for patients with hematologic malignancies (blue line) and solid tumors (green line). Data points indicate average screening prevalence per quarter (Q) of each year. Q1, Jan 1-Mar 31; Q2, Apr 1-Jun 30; Q3, Jul 1-Sept 30; Q4, Oct 1-Dec 31. Numbers at top of figure refer to publication of national recommendations and associated reference number, as follows: 1, US Food and Drug Administration; 2, American Association for the Study of Liver Diseases (2007); 3, Centers for Disease Control and Prevention; 4, American Association for the Study of Liver Diseases (2009); 5, National Comprehensive Cancer Network; 6, Institute of Medicine; and 7, American Society of Clinical Oncology.

Discussion

We found that the HBV screening prevalence among new patients receiving chemotherapy at a large US cancer center over the period 2004–2011 was only 16.2%. Of particular concern, the prevalence of HBV screening was low (<19%) even for patients with known HBV risk factors. Over 66% of patients with hematologic malignancies but less than 4% of those with solid tumors were screened. Predictors of HBV screening included having an HBV risk factor and planned rituximab therapy. Interestingly, race/ethnicity was associated with the likelihood of HBV screening for patients with hematologic malignancies but not with solid tumors. Importantly, HBV screening prevalence increased over time and higher rates were sustained after publication of national HBV screening recommendations. In this study of provider-driven screening, 8.3% of screened patients had a positive HBsAg or anti-HBc test result. The proportion of patients who tested positive for HBV infection increased by over 20% from period 1 to period 3, suggesting that increased screening may lead to increased identification of patients with HBV infection.

The finding that most patients with hematologic malignancies were screened for HBV infection whereas most patients with known HBV risk factors were not, together with the finding that most patients who received rituximab, a known risk factor for reactivation, were screened, suggests that oncologists are more aware of the risk factors for HBV reactivation than they are of the risk factors for HBV infection. This may have reflected the effect of the FDA letters, package inserts, and recommendations as well as publications in the oncology literature about HBV reactivation associated with rituximab treatment. These data indicate that future educational efforts on risk factors for HBV infection for oncology providers might increase HBV screening.

For patients with hematologic malignancies, the prevalence of HBV screening increased dramatically during period 1, which included the FDA letter. This increase may be related to the high risk (nearly 50%) of reactivation [16] and frequent reports of reactivation among patients with hematologic malignancies [1719] and to the frequent reports of reactivation among patients receiving rituximab [2024]. The further increase in screening prevalence during periods 2 and 3 was likely due to the emphasis in national recommendations on the risk of HBV reactivation in these patients.

For patients with solid tumors, odds of screening increased over all 3 time periods; however, the vast majority (96%) of patients were not screened. The low rate of HBV screening among patients with solid tumors is concerning because of previous reports of reactivation and related delays in chemotherapy and increases in mortality in patients with breast cancer [2527], glioblastoma [28], germ cell tumors [27], and cancers of the lung, colon, and stomach [27, 2931]. Indeed, the risk of reactivation among patients with solid tumors is estimated to be approximately 15% [27]; however, these data were derived in an HBV-endemic area. Therefore, studies are needed to define risks and to determine predictors of reactivation for US patients with solid tumors.

Most of the national recommendations [1, 2, 47] call for prechemotherapy HBV screening in patients with high risk of HBV infection. Although the overall screening prevalence among patients with HBV risk factors was low, the prevalence increased over time, and having an HBV risk factor predicted screening. However, since previous studies have shown that screening based on risk factors alone would miss up to 45% to 65% of patients who actually had HBV infection, [32, 33] future research is warranted to better understand the efficacy of risk-based screening.

Our study’s screening rate is lower than that in previous studies, which have described rates of adherence to cancer-related guidelines ranging from 27% to 97%, [3441] although it is possible that our screening rates may have underestimated the actual rate since we could not verify HBV screening performed before registration at MD Anderson. Reasons for noncompliance with HBV screening guidelines may have included patient characteristics such as age [34, 38] and stage of disease [34], physician attitudes towards guidelines [42], and education about guidelines [43]. One study [44] found that physicians’ lack of awareness of and lack of agreement with guidelines were potential barriers to adherence. A previous study by In et al. [45] reported a higher variation in surgical cancer care when guidelines were based on low levels of evidence or expert opinion. Future research providing high levels of evidence will be necessary to improve adherence to HBV screening.

We found that rituximab was a predictor of screening for all patients, especially those with solid tumors. Rituximab is a monoclonal antibody against CD20+ that causes severe B-cell depletion [46, 47] and facilitates uncontrolled replication of HBV. However, besides rituximab, many other chemotherapy drugs [2528, 4854] have been associated with HBV reactivation. Future studies focusing on mechanisms by which certain chemotherapy drugs may cause reactivation will help shape future evidence-based screening strategies.

Interestingly, whereas race/ethnicity did not predict HBV screening among patients with solid tumors, among patients with hematologic malignancies, Black patients had lower odds of screening than White patients. This is concerning because previous population-based studies have shown that the prevalence of HBV infection (current and past) is higher among Black than White adults (9.6% vs. 2.3%, P < 0.001) [55]. Perhaps physicians are unaware of the higher HBV risk in Black patients. We were surprised that Asian race did not predict HBV screening even though the prevalence of chronic HBV infection in this group may be as high as 20% [5557]. Failure to screen Asian patients may have reflected lack of awareness by physicians of HBV risk factors [58, 59]. In addition, we were surprised that patients with solid tumors who resided outside the US had lower odds of HBV screening, although it is possible that they were screened in their home countries.

We found substantial numbers of patients who had a negative HBsAg test result but a positive anti-HBc test result. Such patients may have occult HBV infection, as underscored by the high risk (78%) of HBV transmission in recipients transplanted with livers from donors with isolated anti-HBc positivity as compared to donors who were anti-HBc negative (0.05%) [60]. It is possible that isolated anti-HBc may represent false-positive test result among populations with a low prevalence of HBV infection. However, reactivation has been reported in patients who are HBsAg negative but anti-HBc positive during chemotherapy particularly if the regimen includes rituximab [21, 24, 61]. The ASCO PCO [7] recommends anti-HBc testing in some populations—e.g., patients with hematologic malignancies—since the risk of reactivation has been reported to be 10% among patients with hematologic malignancies with isolated anti-HBc [62]. The CDC recommends HBV screening using 3 HBV serology tests. We found that anti-HBs was rarely tested during our study period.

The strengths of our study include the large and heterogeneous patient population and the focus on actual rather than recalled HBV screening practice. Previous survey studies estimated 38%-80% of physicians screen patients before chemotherapy [1113]; however, those results may inaccurately describe screening patterns since surveys record self-reporting of screening practice and not actual screening of individual patients. Our examination of physicians’ actual screening behavior at the level of individual patients avoided recall bias or subconscious attempts to report what should be done rather than what was actually done.

The main limitation of our study is its retrospective design and use of administrative databases, which prevented us from fully assessing HBV history and HBV risk factors. Patients may have received chemotherapy before their first chemotherapy administration at MD Anderson. Also, we excluded oral chemotherapy because we could not accurately access dispensing records outside of MD Anderson, but some oral chemotherapy could cause HBV. Patients’ race/ethnicity was self-described or assigned by referring clinics and may be incorrect. Another limitation is that we were not able to accurately determine prevalence of reactivation since not all patients who received chemotherapy were screened for HBV infection. This single-institution experience may not be generalizable to other settings, and our data cannot be generalized to patients who receive care in clinical trials as such patients were excluded. We did not explore socioeconomic factors such as income and educational level because this information is not part of our institutional Tumor Registry database. Most of our patients at MD Anderson have health insurance, and these plans are expected to pay for HBV screening tests. Finally, the last time period in our study was relatively short, limiting our ability to assess the full impact of the national recommendations. Nevertheless, our study provides valuable data from a large US academic cancer center with no changes in institutional policies regarding HBV screening during the study period.

Conclusions

In conclusion, we found that the prevalence of HBV screening before chemotherapy among new patients receiving chemotherapy at a large US cancer center during 2004–2011 was only 16.2% overall but increased over time. The vast majority of patients with solid tumors, even those with risk factors for HBV infection, remained unscreened. Future research is needed to explore risks and predictors of reactivation with chemotherapy for US patients to develop evidence-based screening guidelines. Once these are available, educational efforts should be developed to increase oncology medical providers’ awareness of the importance of HBV screening and prophylaxis to prevent reactivation due to chemotherapy.