Unexpectedly frequent hepatitis B reactivation by chemoradiation in postgastrectomy patients


  • Jason Chia-Hsien Cheng M.D., M.S.,

    Corresponding author
    1. Department of Radiation Oncology, Koo Foundation Sun Yat-Sen Cancer Center, Taipei, Taiwan
    2. Cancer Research Center, National Taiwan University Hospital and College of Medicine, Taipei, Taiwan
    3. Department of Oncology, National Taiwan University Hospital and College of Medicine, Taipei, Taiwan
    • Division of Radiation Oncology, Department of Oncology, National Taiwan University Hospital, No. 7 Chung-Shan South Road, Taipei 100, Taiwan
    Search for more papers by this author
    • Fax: (011) 886 223711174

  • Mei-Ching Liu M.D.,

    1. Department of Medical Oncology, Koo Foundation Sun Yat-Sen Cancer Center, Taipei, Taiwan
    Search for more papers by this author
  • Stella Y. Tsai M.D.,

    1. Department of Radiation Oncology, Koo Foundation Sun Yat-Sen Cancer Center, Taipei, Taiwan
    Search for more papers by this author
  • Wei-Tse Fang M.D.,

    1. Department of Medical Oncology, Koo Foundation Sun Yat-Sen Cancer Center, Taipei, Taiwan
    Search for more papers by this author
  • James Jer-Min Jian M.D.,

    1. Department of Radiation Oncology, Koo Foundation Sun Yat-Sen Cancer Center, Taipei, Taiwan
    Search for more papers by this author
  • Juei-Low Sung M.D.

    1. Department of Internal Medicine, Koo Foundation Sun Yat-Sen Cancer Center, Taipei, Taiwan
    Search for more papers by this author

  • Presented in part at the 44th Annual Meeting of the American Society for Therapeutic Radiology and Oncology, New Orleans, Louisiana, October 6–9, 2002.



Postgastrectomy patients undergoing chemoradiation risk chemoradiation-induced liver disease (CRILD). The objectives of this study were to investigate dosimetric implications and assess biologic susceptibility to CRILD in these patients.


Sixty-two patients with Stage IB–IV gastric/gastroesophageal adenocarcinoma without metastases underwent radical total/subtotal gastrectomy; regional lymph node dissection; and postoperative, adjuvant, concomitant chemoradiotherapy (CCRT). Among these, 8 patients developed CRILD (defined as Grade 3–4 liver toxicity), and 11 patients were chronic hepatitis B virus (HBV) carriers (HBV+). Chemotherapy consisted of 1 cycle of etoposide, leucovorin, and 5-fluorouracil (ELF); followed by 5 weekly high doses of 5-fluorouracil (2000–2600 mg/m2) and leucovorin concurrent with radiotherapy (median dose, 45 grays [Gy] to the tumor bed/regional lymphatics); followed by 3 cycles of ELF separated by a 21-day interval. Patients were followed for ≥ 4 months after CCRT. Patient-related and dosimetric factors were correlated with CRILD.


HBV+ status was the only independent factor associated with CRILD. HBV+ patients had a higher CRILD incidence (6 of 11 patients vs. 2 of 51 patients; P < 0.001). HBV-negative patients with CRILD were recipients of a higher mean liver dose (MLD) (23.8 Gy vs. 15.2 Gy; P = 0.009) and a higher volume fraction of liver that received > 30 Gy (36.5% vs. 19.7%; P = 0.009) compared with noncarriers without CRILD, but no MLD difference was found between HBV+ patients with or without CRILD. Moreover, in four of six carriers with CRILD, HBV infection was reactivated during CRILD. Two of the toxicities were fatal.


HBV carriers had a higher incidence of CRILD after postgastrectomy CCRT, probably related to HBV reactivation. Dosimetric parameters modulated the risk of CRILD in noncarriers, but not in carriers. These factors deserve attention in CRILD/HBV+ patients, and the underlying pathogenesis warrants investigation. Cancer 2004. © 2004 American Cancer Society.

Among the treatment modalities for gastric carcinoma, radical surgery remains the potentially most curative treatment for 20–85% of patients.1 Recently, published data have established further the benefits of postoperative, adjuvant, concomitant chemoradiotherapy (CCRT) for disease control and survival in patients at high risk for recurrence.2 The side effects from an intensive combined-modality treatment deserve special attention and demand the use of caution in care giving.2, 3 Most treatment-related, adverse reactions are hematologic or gastrointestinal problems, whereas hepatic toxicity is described less frequently.2, 3 Studies on predisposing factors and the pathophysiology of radiation-induced liver disease (in which the radiation is to part of the liver) have been reported.4–6 In addition, chemotherapeutic regimens sometimes have been associated with hepatic toxicity, such as toxicity related to reactivation of chronic viral hepatitis in patients who were carriers of hepatitis B virus (HBV).7 In the current study, we found an unexpectedly frequent incidence of hepatic toxicity in HBV carriers with gastric carcinoma who underwent postgastrectomy adjuvant CCRT. We also investigated biologic correlations of postgastrectomy adjuvant CCRT with hepatic toxicity, dosimetric implications, and a possible role in the pathogenesis of hepatic toxicity.


From August, 1993 through December, 2001, 71 patients at our institution were diagnosed with high-risk, Stage IB–IV M0 adenocarcinoma of the stomach or gastric cardia and underwent radical total or subtotal gastrectomy; regional lymph node dissection; and received the in-house, Institutional Review Board-approved regimen of postoperative adjuvant CCRT. Among them, 62 patients with complete dose-volume radiation therapy data, adequate follow-up of at least 4 months after CCRT, and serum virology tests for HBV antigen were included in the current study. Excluded were 5 patients without serum virology results (1 of whom had a follow-up interval of < 4 months) and 4 patients with deviations from the protocol (i.e., bolus infusion of weekly 5-fluorouracil < 500 mg/m2 rather than the designed, 24-hour, continuous infusion of 2000 mg/m2). Eleven of 62 patients were chronic HBV carriers (defined as positive serum hepatitis B surface antigen > 4 ng/mL according to a microparticle enzyme immunoassay). Three patients, other than the 11 HBV carriers, were seropositive for the anti-HCV antibody. None of the patients in this study had biochemical, hematologic, or imaging evidence of cirrhosis in preoperative staging work-ups. All patients had hepatic function within normal limits before CCRT. Patient characteristics are shown in Table 1.

Table 1. Patient Characteristics (n = 62)
Patient characteristic 
  1. AJCC: American Joint Committee on Cancer; CCRT: concomitant chemoradiotherapy; SD: standard deviation; Gy: grays; V30: the volume fraction of the liver that received a radiation dose > 30 Gy.

Categoric variables:No. of patients (%)
  Male40 (65)
  Female22 (35)
 AJCC stage 
  Stage IB 4 (6)
  Stage II14 (23)
  Stage IIIA21 (34)
  Stage IIIB 8 (13)
  Stage IV15 (24)
 Tumor classification 
  T1 2 (3)
  T217 (27)
  T337 (60)
  T4 6 (10)
 Lymph node classification 
  N0 8 (13)
  N128 (45)
  N216 (26)
  N310 (16)
 Chronic hepatitis B virus carrier 
  Yes11 (18)
  No51 (82)
 Pre-CCRT chemotherapy 
  Yes58 (94)
  No 4 (6)
 Post-CCRT chemotherapy 
  Yes55 (89)
  No 7 (11)
 Cycles of chemotherapy during CCRT 
  1 3 (5)
  2 2 (3)
  3 2 (3)
  414 (23)
  536 (58)
  6 5 (8)
Continuous variables: Mean ± SD (range) 
 Age (yrs)52.9 ± 12.5 (21.5–80.9)
 Isocenter dose (Gy)45.4 ± 1.9 (41.4–57.10)
 Normal liver volume (mL)1157.6 ± 242.0 (707–1791)
 Mean dose to liver (Gy)15.7 ± 4.5 (6.8–24.9)
 V30 (%)20.6 ± 9.3 (1.0–43.0)

Radiation treatment usually started 6–8 weeks after gastrectomy. The radiotherapy protocol for postgastrectomy adjuvant treatment was similar to that used in the Intergroup 0116 Gastric Surgical Adjuvant Trial,2 including 41.4–57.1 grays (Gy) (median, 45 Gy) in daily fractions of 1.8–2.0 Gy, with 6-megavolt (MV) (mainly in the anterior-posterior field) or 18-MV photon beams directed to the tumor bed, the regional lymph nodes, and 1.5–20 cm beyond the proximal and distal margins of resection. Two-field (anterior-posterior/posterior-anterior), three-field (anterior-posterior/bilateral opposed with wedge pairs), or four-field box-beam designs were used for the treatment. The length of horizontal axis was usually 15–18 cm in anterior-posterior/posterior-anterior (AP/PA) fields and 12–15 cm in opposing lateral fields. The length of the longitudinal axis was usually 15–20 cm in these fields. Radiation portals were designed with a computed tomography (CT)-based treatment planning system (FOCUS system; CMS Company, St. Louis, MO), which generated the complete dose-volume histograms (DVHs) of targets and organs at risk. Doses were limited, so that < 60% of the hepatic volume was exposed to > 30 Gy of radiation. The equivalent of at least two-thirds of the kidney was spared from the field of radiation. (At least three-fourths of the kidney should be excluded when doses of 20 Gy are used.) No portion of the heart representing 30% of the cardiac volume received > 40 Gy of radiation. A typical DVH of liver for patients in this study is shown in Figure 1.

Figure 1.

A typical liver dose-volume histogram for patients undergoing radical subtotal gastrectomy and postoperative adjuvant concomitant chemoradiotherapy using anterior-posterior/posterior-anterior fields (mean liver dose, 18.2 grays). cGy: centigrays.

The chemotherapy protocol included a weekly cycle of high-dose 5-fluorouracil (median, 2000 mg/m2; range, 1400–2600 mg/m2) and leucovorin (300 mg/m2) as a continuous, intravenous infusion for 24 hours concurrent with radiotherapy. Concomitant chemotherapy was given in a median number of 5 cycles (range, 1–6 cycles). Due to severe bone marrow suppression, the dose of concomitant 5-fluorouracil was reduced to < 2000 mg/m2 (1400 mg/m2) in 1 patient. Fifty-eight patients (94%) received 1 cycle of chemotherapy with etoposide (120 mg/m2 per day), leucovorin (300 mg/m2 per day), and 5-fluorouracil (500 mg/m2 per day) (the ELF regimen) for 3 days; this regimen was given 3 weeks before CCRT. Fifty-five patients (89%) also received post-CCRT chemotherapy. Among them, 51 patients had a median of 3 cycles (range, 1–3 cycles) of post-CCRT chemotherapy using ELF regimens in a 21-day interval starting 3 weeks after CCRT. Four patients maintained the post-CCRT chemotherapy using weekly, high-dose 5-fluorouracil and leucovorin for a median of 9 cycles (range, 5–11 cycles).

Patients were monitored weekly during CCRT, with history-taking, physical examination, blood counts, and blood chemistries, including liver function tests, and were they were followed twice each month after CCRT for at least 4 months. Chemoradiation-induced liver disease (CRILD) was defined as Grade ≥ 3 hepatic toxicity, as described in version 2.0 of the National Cancer Institute Common Toxicity Criteria,8 within 4 months after the completion of CCRT. CT scans of the abdomen, liver biopsies, and serum virology tests were obtained for patients with CRILD to exclude the possibility of other causes not related to CCRT.

Patient-related, treatment-related, and dosimetry-related factors were analyzed for their correlation with CRILD. In univariate analysis, a Student t test was used to compare continuous variables between patients with and without CRILD, and a chi-square test or a Fisher exact test was used to compare categorical variables. Logistic regression analysis was used to investigate multivariable impact on the occurrence of CRILD. P values < 0.05 for 2-sided tests were considered statistically significant. Three dosimetric parameters, the percent volume of normal liver that received a radiation dose > 30 Gy (V30), the mean radiation dose to the liver (MLD), and the Lyman normal tissue complication probability (NTCP),9 were obtained from the DVHs. The MLD was calculated by dividing the sum of the doses to all voxels of normal liver by the number of voxels. The 3 Lyman NTCP model parameters, TD50(1), n, and m, were adjusted to best fit the true condition (with or without CRILD) in each patient using the maximal-likelihood method.10 Ninety-five percent confidence intervals (95% CI) of the parameters were determined by the profile-likelihood method.11 The following Lyman NTCP model was used:

equation image

Data were fit to the Lyman NTCP model, which describes the probability of a complication after uniform radiation of a fractional volume of normal tissue (v) to a dose (D), assuming a sigmoid dose-response relation with no threshold. TD50(v) represents the tolerances doses associated with a 50% chance of complications for uniform partial liver irradiation, where TD50(v) is related to the whole liver (v = 1) tolerance through the power law relation: TD50(1) = TD50(v) × Vn.

Vref is the volume of nonneoplastic liver. The parameter n is the volume-effect parameter, which relates the tolerance doses of uniform, whole-organ irradiation to uniform, partial-organ irradiation. The parameter m is the steepness of the dose-response curve at TD50(1). The partial volume-dose–complication risk relation for the liver was displayed graphically in dose-effective volume plots (as a function of NTCP).


Eight patients developed CRILD after the completion of adjuvant CCRT in a median latent period of 29 days (range, 21–74 days). Most patients with CRILD presented with no symptomatic manifestation other than the increased liver enzymes at 1 month after the completion of radiotherapy. The serum HBV DNA level, if available, started rising 2 weeks preceding the increase in liver enzymes. All but 1 patient received the isocenter dose of 45 Gy. Six of 8 patients with CRILD were HBV carriers, and 4 of those patients also had evidence of viral hepatitis B reactivation (serum HBV DNA > 5 pg/mL) at the time of CRILD. The other types of viral hepatitis, including type A, C, and D, were excluded serologically. Although none of 62 patients had biochemical evidence of abnormal hepatic function (transaminase levels < 40 U/L), all 6 carrier patients with CRILD had transaminase levels as high as 500-–500 U/L. Six patients underwent liver biopsy, including four HBV carriers and two noncarriers. The histologic features of the carrier patient biopsies were lymphocytic infiltration in the portal areas, periportal piecemeal necrosis, limiting plate destruction, Kupffer cell hyperplasia, and the presence of apoptotic eosinophilic bodies within the lobules. The noncarrier patient biopsies had similar inflammatory infiltrates but relatively preserved liver plates and lobules. Four HBV carrier patients with CRILD were treated with lamivudine (100 mg per day). One of the two carrier patients with CRILD who were not treated with lamivudine died of hepatic failure. The predominant clinical presentation in all six carrier patients with CRILD was a greater-than-fivefold increase in serum transaminase levels. In contrast, the two noncarrier patients with CRILD presented exclusively with a greater-than-fivefold increase in serum alkaline phosphatase levels but only moderate increases (less than fivefold) in transaminase levels and other liver enzyme levels. The two noncarrier patients with CRILD were asymptomatic clinically, underwent conservative treatment, and were observed closely. Both patients recovered serologically in 2–4 months and had no clinical event related to CRILD. The characteristics of patients with CRILD are shown in Table 2.

Table 2. Characteristics of the Eight Patients with Chemoradiation-Induced Liver Disease
CharacteristicPatient no.
  1. M: male; F: female; AJCC: American Joint Committee on Cancer; HBV hepatitis B virus; Gy: grays; V30: the volume fraction of the normal liver that received a radiation dose > 30 Gy; CT: chemotherapy; CCRT: concomitant chemoradiotherapy: CRILD: chemoradiation-induced liver disease; R: recovery; F: fatality.

Age (yrs)5141405572584360
HBV carrierYesYesYesYesYesYesNoNo
Volume of liver (mL)7541791884126390312221283994
Isocenter dose (Gy)50.445.
Mean radiation dose to the liver (Gy)10.420.722.912.918.513.122.724.9
V30 (%)83537242473043
Cycles of concomitant CT65555555
Pre-CCRT CTNoYesYesYesNoYesYesYes
Post-CCRT CTNoYesYesYesYesNoYesNo
Maximal hematologic toxicity grade during CCRT12223212
Latent period of CRILD (days)2626632974292155
HBV DNA (pg/mL)3.31485640333326
Lamivudine treatmentNoYesYesYesYesNo

In univariate analysis, status as an HBV carrier was the only factor significantly associated with CRILD (6 of 11 patients vs. 2 of 51 patients; P < 0.001). Other factors that did not reach statistical significance included age (P = 0.92), gender (P = 0.70), T classification (P = 1.0), lymph node status (P = 0.71), American Joint Committee on Cancer disease stage (P = 1.0), pre-CCRT chemotherapy (P = 0.08), maximal hematologic toxicity during CCRT (P = 0.26), normal liver volume (P = 0.80), isocenter dose (P = 0.43), V30 (P = 0.08), and MLD (P = 0.09) (Table 3). In multivariate analysis with significant or borderline significant factors from the univariate test, including positive HBV carrier status, pre-CCRT chemotherapy, V30, and MLD, status as an HBV carrier was the only independent factor significantly associated with CRILD (odds ratio, 36.7; 95% CI, 5.5–245.0; P = 0.0002).

Table 3. Univariate Analysis of Patient-Related, Treatment-Related, and Dosimetric Factors Associated with Chemoradiation-Induced Liver Disease
VariableCRILD (n = 8)No CRILD (n = 54)P value
  1. CRILD: chemoradiation-induced liver disease; SD: standard deviation; HBV: hepatitis B virus; AJCC: American Joint Committee on Cancer; CCRT: concomitant chemoradiation therapy; Gy: grays; V30: volume fraction of normal liver that received a radiation dose > 30 Gy.

Gender (male/female)6/234/200.70
Mean age ± SD (yrs)52.4 ± 10.952.9 ± 12.90.92
HBV carrier (yes/no)6/25/49< 0.001
AJCC stage (Stage IB–II/Stage IIIA–IV)2/616/381.0
T stage (T1–2 T3–4)2/617/371.0
Lymph node classification (N0–1/N2–3)4/432/220.77
Pre-CCRT chemotherapy (yes/no)6/252/20.08
Maximal hematologic toxicity grade during CCRT (Grade 0–1/Grade > 1)2/627/270.26
Mean ± SD normal liver volume (mL)1136.8 ± 329.11160.7 ± 230.30.80
Mean ± SD isocenter dose (Gy)45.9 ± 1.945.3 ± 1.90.43
Mean ± SD V30 (%)26.0 ± 13.119.8 ± 8.50.08
Mean ± SD radiation dose to the liver (Gy)18.3 ± 5.415.4 ± 4.30.09

In the subgroup analysis of patients who were not HBV carriers, patients with CRILD had significantly higher MLD (23.8 Gy vs. 15.2 Gy; P = 0.009) and V30 (36.5% vs. 19.7%; P = 0.009) than those without CRILD. Among HBV carriers, patients with and without CRILD had similar MLD (16.4 Gy vs. 17.2 Gy; P = 0.74) and V30 (22.5% vs. 20.6%; P = 0.79) (Table 4). None of the other factors showed statistically significant differences between patients with and without CRILD in the two subgroups. For noncarrier patients, the parameters of the Lyman NTCP model, TD50(1), m, and n, were 21.5 Gy (95% CI, 19.4–24.5 Gy), 0.09 (95% CI, 0.03–0.24), and 1.99 (95% CI, 1.20–7.15), respectively; for HBV carriers, the same parameters were 20.5 Gy (95% CI, 5.2 Gy to > 100 Gy), 6.88 (95% CI, 0.89 to > 2.00), and 0.11 (95% CI, < 0.01 to > 2.00), respectively (Table 4).

Table 4. Dosimetric Comparison between Patients Who Had Gastric Carcinoma with and without Chemoradiation-Induced Liver Disease and Patients Who Had Hepatocellular Carcinoma with Radiation-Induced Liver Disease
SubgroupNo. of patientsCRILD or RILDTD50(1) (95% CI) in GyParameter m (95% CI)Parameter n (95% CI)Mean ± SD RT dose to the liver (Gy)P valueMean ± SD V30 (%)P value
  1. CRILD: chemoradiation-induced liver disease; RILD: radiation-induced liver disease; TD50(1): the tolerance dose of uniform partial liver irradiation associated with a 50% chance of complications; Gy: grays; 95% CI: 95% confidence interval; Parameter m: steepness of the dose-response curve at TD50(1); Parameter n: volume-effect parameter; SD: standard deviation; RT: radiotherapy; V30: volume fraction of normal liver that received a radiation dose > 30 Gy; HBV: hepatitis B virus; HCC: hepatocellular carcinoma.

Non-HBV carriers         
 With gastric carcinoma49Negative21.5 (19.4–24.5)0.09 (0.03–0.24)1.99 (1.20–7.15)15.2 ± 4.40.00919.7 ± 8.50.009
 With gastric carcinoma2Positive   23.8 ± 1.6 36.5 ± 9.2 
HBV carriers         
 With gastric carcinoma5Negative20.5 (5.2 to > 100.0)6.88 (0.89 to > 2.00)0.11 (< 0.01 to > 2.00)17.2 ± 2.40.7420.6 ± 9.70.79
 With gastric carcinoma6Positive   16.4 ± 5.0 22.5 ± 12.8 
 With HCC16/65Positive50.2 (42.7–61.3)0.39 (0.26–0.71)0.24 (0.12–0.44)23.0 ± 8.20.08538.6 ± 16.20.04

The hepatic tolerance of 6 carrier patients with CRILD in this study was compared with the tolerance of 16 patients who developed radiation-induced liver disease from a separate group of 65 patients who were HBV carriers and underwent radiotherapy alone for their hepatocellular carcinoma (HCC). The carrier patients with gastric carcinoma and CRILD had borderline significantly lower MLD (16.4 Gy vs. 23.0 Gy; P = 0.085) and significantly lower V30 (22.5% vs. 38.6%; P = 0.04) (Table 4).


The recently updated and encouraging data from the Gastric Surgical Adjuvant Trial Intergroup 0116 study showed an increase in recurrence-free and overall survival stemming from the addition of adjuvant CCRT for patients with resected high-risk, Stage IB–IV M0 adenocarcinoma of the stomach or gastroesophageal junction.2 In that report, treatment-related toxicity was mainly hematologic or gastrointestinal. Four patients (1%) had Grade ≥ 3 hepatic toxicity, which resembled the toxicity associated with CRILD in our study. However, the details of these events were not described, and the chronic viral hepatitis status of their patients was not stated. In the Intergroup Trial, the chemotherapy started with 3-day or 5-day 5-fluorouracil at doses of 400–425 mg/m2 per day and leucovorin at a dose of 20 mg/m2 per day during, before, and after CCRT, in 4-week intervals. In contrast, we used weekly cycles of high-dose 5-fluorouracil at doses of 2000–2600 mg/m2 and leucovorin at a dose of 300 mg/m2 given over a 24-hour period during CCRT and the ELF regimen before and after CCRT. The chemotherapy in our study may have been more intense than that used in the Intergroup Trial. Most reported studies used continuous infusion of 5-fluorouracil 200–500 mg/m2 per day concurrently with radiotherapy for patients with gastrointestinal or hepatobiliary malignancies. Seydel et al.12 reported 1 patient with pancreatic carcinoma who was dying of hepatic failure 3 months after twice-daily, hyperfractionated radiotherapy (50.4-Gy, 1.2-Gy, and 42.0-Gy fractions) concomitantly with 5-fluorouracil 350 mg/m2 per day for 3 days in the first and last weeks of radiotherapy. Those authors did not specify the dose-volume criteria for the liver in their protocol, nor did they investigate the underlying cause of this complication. Evans et al.13 reported 2 patients with pancreatic carcinoma who died of treatment-related causes after receiving concurrent 5-fluorouracil at a dose of 300 mg/m2 per day and conventional radiotherapy (50.4 Gy) with prophylactic, whole-liver irradiation (23.4 Gy). That trial was terminated because of its severe morbidity. Abrams et al.14 also reported a possible case of radiation hepatitis after concurrent 5-fluorouracil at a dose of 200 mg/m2 per day during a radiotherapy course, which consisted of whole-liver irradiation (23.4–27.0 Gy) and pancreatic tumor/lymphatics irradiation (50.4–57.6 Gy). No severe liver toxicity was described in the other trials that used either once-daily radiotherapy along with concurrent 5-fluorouracil of conventional dosage or partial-liver irradiation alone.15–17

There have been several studies using higher doses of 5-fluorouracil (≥ 1000 mg/m2 per day) with or without concomitant radiotherapy. Komaki et al.18 reported 1 case of fatal hepatic toxicity after continuous-infusion 5-fluorouracil at a dose of 1000 mg/m2 per day for 5 days in a monthly cycle during radiotherapy, which included pancreatic irradiation (61.2 Gy) and prophylactic hepatic irradiation (23.4 Gy). The German series showed no liver toxicity for patients with pancreatic carcinoma who underwent conventional radiotherapy (55.8 Gy) and chemotherapy with 5-fluorouracil at a dose of 1000 mg/m2 per day for 5 days with mitomycin C in a 4-week cycle.19 The other reports did not describe any liver toxicity when using even higher doses of 5-fluorouracil (2000–2600 mg/m2 per day) but found no concomitant radiotherapy in patients with advanced gastric carcinoma.20, 21 None of those reports mentioned information regarding preexisting liver disease in their patients, such as chronic viral hepatitis. In our study, the adjuvant treatment protocol included weekly, high-dose 5-fluorouracil (2000–2600 mg/m2 day) and leucovorin concurrent with conventional radiotherapy (45.0-Gy, 1.8-Gy, and 25.0-Gy fractions) to the tumor bed/lymphatics. Even with normal fractionation and without whole-liver irradiation, 8 of our patients (13%) still showed Grade ≥ 3 liver toxicity. Whether the adverse effect was from the chemotherapy or from partial hepatic irradiation remains unsettled and deserves further investigation.

The definition of CRILD in this study was Grade ≥ 3 hepatic toxicity, which was evaluated by clinical signs and liver function tests. Both CRILD and radiation-induced liver disease, as defined in the radiation series,22, 23 shared the same criteria based on the clinical presentation rather than the underlying etiology of hepatic damage. Our previous report indicated that our patients with HCC and radiation-induced liver disease after conformal radiotherapy mainly presented with elevated levels of serum transaminases rather than abnormally increased alkaline phosphatase levels and the commonly described ascites.24 The striking difference may be from the high prevalence in our patients of chronic viral hepatitis, which is associated with more significant radiation-induced injury to hepatocytes than to bile ducts. In addition, the underlying etiology of radiation-induced liver disease in our series of patients with HCC may have been the reactivation of viral hepatitis, especially in the most recent patients with available serum virology results. In the current study, six of the eight patients with CRILD were HBV carriers. Four of the six carriers also had serologic evidence of viral hepatitis reactivation during CRILD. An increased level of HBV DNA usually precedes the appearance of abnormal liver function and clinical signs and symptoms.25 In the current study, the timing of the serum virology tests may have confounded the results of the two carrier patients with CRILD but without increased HBV DNA level. Therefore, our unique findings partly may imply viral hepatitis reactivated by radiotherapy or CCRT as a possible pathogenesis of radiation-induced liver disease or CRILD for patients who are HBV carriers. Furthermore, the dosimetric comparison between carrier patients with CRILD in this study and patients with HCC and radiation-induced liver disease showed that hepatic tolerance was even lower to CCRT (16.4 Gy) than to irradiation alone (23.0 Gy). It is likely that chemotherapy played some role in triggering or enhancing radiation damage to the liver.

Reactivation of hepatitis B may occur sporadically, with or without specific predisposing factors.26 Chemotherapy has been one of the most frequently reported factors associated with reactivation of viral hepatitis B.7, 27, 28 Yeo et al. reported the largest series with 15 of 78 carrier patients developing reactivation of hepatitis B after chemotherapy (7). Carrier patients who were negative for hepatitis B e (HBe) antigen and positive for anti-HBe with a nucleotide 1896 mutation were more likely to have reactivation during chemotherapy than those with wild type virus.29 The antiviral nucleoside analogue, lamivudine, was identified as effective in suppressing viral replication and reactivation of hepatitis.30 Similarly, but more important, > 50% of our carrier patients in the current study developed CRILD, and most of them recovered after lamivudine therapy. The success of lamivudine was in the suppression of viral hepatitis B reactivation, which was documented serologically in four of the six carrier patients during CRILD.

Smalley et al.3 discussed the treatment implementation of post-gastrectomy adjuvant CCRT in another report. Parallel-opposed AP-PA fields were preferred to cover the tumor bed and the possible extent of disease spread adequately. In addition, lateral portals, if used, should be limited to 20 Gy to avoid the exposure of a significant hepatic volume. Technically, the protocol of the Intergroup 0116 Trial only limited < 60% of the liver from receiving 30 Gy of radiation (V30).2V30 has been used as a volume threshold of potential hepatic injury by irradiation.31 Other dosimetric parameters, including MLD and the Lyman NTCP model, proved more useful in estimating the risk of radiation-induced liver disease for patients with liver tumors receiving local radiotherapy.5, 6 Our previous study demonstrated the effectiveness of the NTCP model and disclosed a significant loss of hepatic tolerance to irradiation in patients with HCC, a large proportion of whom had chronic viral hepatitis.6 Our data showed significantly higher MLD and V30 values for noncarrier patients with CRILD compared with noncarriers without CRILD. The estimates of NTCP parameters in the noncarrier subgroup also revealed a large volume effect, indicating that the commonly accepted concept of parallel organ characteristics still applies to the liver.5 Although the results in our current series revealed that the dosimetric parameters sustained their effectiveness in distinguishing between noncarrier patients with and without CRILD, dosimetric analysis was no longer useful in assessing risk of CRILD in carrier patients. There were probably too few carrier patients to demonstrate any dosimetric correlation with CRILD or to perform the NTCP parameterization meaningfully. However, the findings may support in part the hypothesis of a different pathogenesis of CRILD, such as viral hepatitis reactivation, in carrier patients. The impact of biologic factors (e.g., HBV) on the development of CRILD may be more important than the impact of dosimetric factors in these carrier patients. Conversely, because trends are evident for the increased risk with the dosimetric parameters, it is possible that dosimetric factors could become significant when more patients are enrolled. The liver toxicity observed, the patient population, and the treatment in this series differed from previous series describing radiation-induced liver toxicity, influencing analysis of the dose-volume–complication relation and possibly explaining the differences between this NTCP fit and other NTCP analyses.

In conclusion, HBV carriers had a significantly higher risk of CRILD after postgastrectomy CCRT. Dosimetric factors of MLD and V30 were effective in differentiating noncarrier patients from patients without CRILD, but they were not effective in differentiating carriers. Reactivation of viral hepatitis B was the underlying etiology in most carriers with CRILD. Carriers with gastric carcinoma had a lower tolerance to CCRT compared with the tolerance of carriers with HCC to irradiation alone. The findings justify an investigation of the exact pathogenesis of CRILD in patients who are HBV carrier. This issue deserves special attention, and carriers undergoing CCRT should be observed closely because of the high probability of liver toxicity.