PDF
Citation
Print
Introduction
Hepatoblastoma, the most common primary hepatic malignancy in children, has an estimated annual incidence of approximately 1.5 cases per million [1]. Treatment includes chemotherapy, surgical resection, and liver transplantation. These approaches have undergone advancements, improving survival outcomes [2]. The International Childhood Liver Tumors Strategy Group typically administers neoadjuvant chemotherapy to hepatoblastoma patients, followed by delayed surgery, with the advantage of reducing tumor size and down-staging in most cases [3].
Several studies have evaluated the prognosis of hepatoblastoma and guided clinical management. Prognostic factors include the PRE-Treatment Extent of tumor (PRETEXT) system, the Children’s Hepatic Tumors International Collaboration-Hepatoblastoma risk stratification (CHIC-HS), the measurement of α-fetoprotein (AFP) levels, and histological subtypes [4,5]. Previous studies have reported that higher PRETEXT group classifications, as well as positive PRETEXT annotation factors P (portal vein), F (multifocality), and M (distant metastases), and either a low (< 100 ng/mL) or a very high (> 106 ng/mL) level of AFP at diagnosis, are associated with unfavorable outcomes in hepatoblastoma patients [4,6]. The CHIC-HS has been introduced as a new risk stratification system for hepatoblastoma patients by collaborating with four major international liver groups [5]. It can be applied during initial diagnosis based on age, PRETEXT system, AFP level, and tumor resectability. Previous studies validated the prognostic effect of the CHIC-HS system in the Asian pediatric population, underscoring its significance as a predictor of event-free survival (EFS) [7,8].
As neoadjuvant chemotherapy has been widely used in the treatment of hepatoblastoma, post-treatment evaluation of the disease is crucial for patient management. The PRETEXT group and PRETEXT annotation factors should be reassessed at each imaging time point. This is called the POST-Treatment Extent of tumor (POSTTEXT) system. It evaluates the tumor extent response to neoadjuvant chemotherapy and provides information about post-treatment resectability. Due to the substantial size of the hepatoblastoma at the time of diagnosis, there is a considerable risk of misevaluation of the PRETEXT system [3]. Also, with remarkable advancements in neoadjuvant chemotherapy, previously deemed unresectable tumors have substantially reduced size, enabling complete resection [3]. In this regard, the accurate evaluation of the POSTTEXT system has gained importance, surpassing reliance on the PRETEXT system alone.
Our study aims to assess the prognostic value of the POSTTEXT system and clinical factors after neoadjuvant chemotherapy in hepatoblastoma patients and evaluate the benefits of post-treatment imaging and clinical factors after neoadjuvant chemotherapy in conjunction with the CHIC-HS system.
Materials and Methods
This study followed Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [9].
1. Patients
A retrospective database search was conducted at a tertiary referral center between March 2006 and March 2022. The study period was from 2006 because liver transplantation was introduced for hepatoblastoma patients at our center that year [2]. Inclusion criteria: (1) diagnosis of hepatoblastoma histopathologically; (2) age under 18 years; (3) underwent neoadjuvant chemotherapy at least four times; (4) abdominal computed tomography (CT) or magnetic resonance imaging (MRI) at the time of diagnosis and after the fourth cycle of neoadjuvant chemotherapy; and (5) accessible laboratory and follow-up electronic medical records. Exclusion criteria: (1) poor imaging quality hindering PRETEXT and POSTTEXT staging and (2) patients lost to follow-up. The fourth cycle of neoadjuvant chemotherapy was chosen because the median number of cycles was four [2]. Additionally, the SIOPEL group recommends four cycles of neoadjuvant chemotherapy for patients who undergo surgical resection [10].
2. PRETEXT and POSTTEXT staging system
Two radiologists (H.M.Y., with 9 years of experience in pediatric radiology, and H.J., with 2 years of experience in radiology) independently assessed the PRETEXT and POSTTEXT staging according to the updated 2017 PRETEXT staging system [11], based on CT or MRI images before treatment and after the fourth cycle of neoadjuvant chemotherapy. They were blinded to the clinical outcome. A PRETEXT and POSTTEXT staging training session was held with 25 cases that were not included in this study. The PRETEXT and POSTTEXT staging indicates the extent of the tumor based on the number of contiguous tumor-free hepatic sections, ranging from PRETEXT I (three consecutive sections free) to PRETEXT IV (no sections free). Details of the annotation factors are explained in Supplementary Material. If the metastatic lesions present at diagnosis were no longer visible post-treatment, POSTTEXT M was considered negative. An aggregate factor, VPEFR, indicating the presence of at least one of the V, P, E, F, or R factors, was also assessed [5]. Any discrepancy between the two radiologists was resolved by consensus after reviewing all available clinical and imaging data. In addition, the largest tumor diameter was measured on CT or MRI at the time of diagnosis and after the fourth cycle of neoadjuvant chemotherapy by one of two radiologists (H.J.).
3. Clinical data collection
Age, sex, and serum AFP levels were collected at diagnosis and after the fourth cycle of neoadjuvant chemotherapy. Patients were classified into very low to high-risk groups according to the CHIC-HS system. “Resectable at diagnosis,” in the lexicon of the CHIC-HS system to differentiate very low to low risk, was retrospectively evaluated by a pediatric surgeon (J.M.N. with 10 years of experience in pediatric liver surgery). The total number and regimen of neoadjuvant chemotherapy and the surgical method were collected.
4. Statistical analysis
The EFS was used as the primary outcome of this study, which is the time from enrollment until an event, including first relapse, disease progression, development of second malignancy, or death for any reason. The Response Evaluation Criteria in Solid Tumors ver. 1.1 criteria was used to evaluate relapse and disease progression [12].
The Kaplan-Meier curves were generated to calculate EFS and overall survival (OS). The OS was defined as the period from diagnosis to death or the most recent follow-up.
Univariable and multivariable Cox regression analyses were conducted to find significant predictors of EFS. Variables that showed a potential for statistical significance (p < 0.05) in the univariable model were included in the multivariable model with a backward elimination process. The POSTTEXT staging, rather than the PRETEXT staging, was primarily included in the multivariable analysis. However, the PRETEXT group was included in the multivariable analysis regardless of the p-value since it is a well-known risk factor associated with EFS [5]. The level of risk associated with each variable was expressed as a hazard ratio (HR), along with a corresponding 95% confidence interval (CI). Additionally, time-dependent receiver operating characteristic (ROC) curves for predicting 5-year EFS were evaluated to compare the predictive power of the CHIC-HS risk system alone and combination with other post-treatment factors. The best cut-off value for post-treatment AFP level and tumor size change (%) were calculated with Uno’s estimator of cumulative area under the curve (AUC) [13] and Youden index.
Inter-reader agreement between the radiologists was analyzed for PRETEXT and POSTTEXT staging using weighted kappa and annotation factors using kappa. Kappa values were interpreted as follows: 0-0.20 (slight agreement), 0.21-0.40 (fair), 0.41-0.60 (moderate), 0.61-0.80 (substantial), and 0.81-1.00 (excellent) [14].
A p < 0.05 was considered statistically significant. The analysis used R software ver. 3.6.1 (R Foundation for Statistical Computing) and MedCalc ver. 22.007 (MedCalc Software).
Results
1. Baseline patient characteristics
Of the 109 patients diagnosed with hepatoblastoma during the study period, the following were excluded: 10 patients without follow-up clinical or imaging data, 15 patients who received less than four times of neoadjuvant chemotherapy, and 11 patients who did not undergo follow-up CT or MRI after four times of chemotherapy. A total of 73 patients were included (Fig. 1). The baseline characteristics of the 73 patients are summarized in Table 1. The mean age of the patients was 2.2±2.7 years, and 39 were male. The mean follow-up period was 3.5 years (range, 0.4 to 16.1 years). Before treatment, 61 patients (n=1/65, 93.8%) had AFP levels of ≥ 1,000 ng/mL. However, after the fourth cycle of chemotherapy, the AFP levels decreased, and only 23 patients (n=23/71, 32.4%) had AFP levels of ≥ 1,000 ng/mL. The size of the tumor was ≥ 10 cm in 55 patients (n=55/73, 75.3%) before treatment, but only 11 patients (15.1%) had a tumor size of ≥ 10 cm after the fourth cycle of chemotherapy. The median percentage changes in AFP levels and tumor size were 99.8% (range, 63.5% to 99.99%) and 44.6% (range, 0% to 90.5%), respectively. There were no patients who developed new metastases or tumor ruptures during neoadjuvant chemotherapy. Among 23 patients who had metastasis at diagnosis (all metastases were in the lung), four patients had no imaging evidence of metastasis after the fourth cycle of chemotherapy, but the other 19 patients had decreased but measurable metastatic lesions after the fourth cycle.
The KM plots for EFS and OS are presented in Fig. 2. The mean EFS was 13.0 years (95% CI, 11.4 to 14.6 years) and the mean OS was 14.5 years (95% CI, 13.2 to 15.8 years). The 1-, 3-, and 5-year EFS rates were 88.9%, 80.3%, and 80.3 %, respectively. The OS rates were 97.1%, 91.8%, and 89.3 % for 1-year, 3-year, and 5-years, respectively.
2. Prognostic factors for predicting event-free survival
In the univariable Cox proportional hazards model, age at diagnosis; PRETEXT annotation factors F, N, and M; AFP level and tumor size after the fourth cycle of neoadjuvant chemotherapy; change ratio of AFP and tumor size; and POSTTEXT annotation factors M were associated with EFS (Table 2).
According to multivariable Cox proportional hazard analysis, AFP level after the fourth cycle of neoadjuvant chemotherapy (HR, 1.233; 95% CI, 1.086 to 1.400, per 10,000 ng/mL; p=0.001); change ratio of tumor size at diagnosis and after the fourth cycle of neoadjuvant chemotherapy (HR, 0.654; 95% CI, 0.448 to 0.955, per 10%; p=0.030); and POSTTEXT annotation factor M (HR, 5.209; 95% CI, 1.639 to 16.553; p=0.005) were significant predictors of EFS (Table 2).
3. Time-dependent ROC analysis for 5-year event-free survival
Table 3 shows AUC values for 5-year EFS obtained by incorporating various combinations of the significant parameters from the multivariable analysis, including POSTTEXT annotation factor M, AFP level, and the change ratio of tumor size after the fourth cycle of neoadjuvant chemotherapy into the well-known hepatoblastoma risk stratification system, CHIC-HS. The predictive power increased when the AFP level after the fourth cycle of neoadjuvant chemotherapy was added to the CHIC-HS (p=0.043). Although the additional combination of POSTTEXT annotation factor M to the former showed a significant increase in the AUC value (p=0.022), the AUC value remained constant at 0.84. Fig. 3 displays the added AUC values of two models, which combined AFP level after the fourth cycle of neoadjuvant chemotherapy and POSTTEXT M positive into the CHIC-HS system. The best cut-off value of the AFP level after the fourth cycle of neoadjuvant chemotherapy and tumor size change ratio for predicting EFS were 1,000 and 28.6% (Fig. 4).
4. Inter-reader agreement
Table 4 shows the inter-reader agreement between the two radiologists for the PRETEXT and POSTTEXT groups and annotation factors. Overall, the POSTTEXT system showed better inter-reader agreement than the PRETEXT system. PRETEXT V and R showed fair and moderate agreement, respectively. All POSTTEXT systems, including POSTTEXT M, showed substantial to near-perfect agreement.
Discussion
This retrospective study demonstrated the significance of various imaging features included in the POSTTEXT system and post-treatment clinical factors as predictors of EFS in patients with hepatoblastoma. Specifically, the AFP level, tumor size change ratio after the fourth cycle of neoadjuvant chemotherapy, and the POSTTEXT annotation factor M emerged as significant predictors. Integrating these factors into the CHIC-HS system significantly enhanced the predictive power. Notably, including AFP level after the fourth cycle of neoadjuvant chemotherapy contributed to a clinically significant improvement in prognosis prediction.
Prior studies have consistently identified AFP levels as a crucial prognostic factor in hepatoblastoma [15,16]. Moreover, a more significant reduction in AFP levels after neoadjuvant chemotherapy was associated with improved prognosis [15]. A significant correlation was observed between the reduction in serum AFP levels after neoadjuvant chemotherapy and a smaller tumor size following the same treatment [15]. Our study aligns with these findings, thereby emphasizing the importance of precise measurement of the AFP levels and changes in tumor size after neoadjuvant chemotherapy as predictors of EFS.
In a previous study analyzing prognostic factors in hepatoblastoma patients, size reduction of less than 25% after chemotherapy, post-chemotherapy metastasis on CT (POSTTEXT M positive), and POSTTEXT-Portal vein involvement (POSTTEXT P positive) were associated with poor outcomes [17]. Similar results were observed in our study except for POSTTEXT P positive. This discrepancy may be attributed to the unavailability of liver transplantation in the institution where the previous research was conducted. For patients with a positive POSTTEXT M, it indicated the presence of a residual tumor even after chemotherapy, probably associated with persistent higher levels of AFP after treatment, potentially leading to poorer outcomes in our study.
Among various POSTTEXT annotation factors, only POSTTEXT M showed prognostic value. Annotation factors V (hepatic vein/inferior vena cava), P, F, and C (caudate lobe involvement) are associated with intrahepatic tumor spread and resectability, and make surgical removal challenging. With advancements in surgical techniques and liver transplantation feasibility, even in cases with positive findings, a favorable prognosis can be anticipated. Annotation factors E (extrahepatic tumor extension), R (tumor rupture), and N (lymph node metastases) are associated with intra-abdominal tumor spread and may also present challenges for complete tumor resection. In our study, these positive factors were relatively rare before treatment, and all those factors turned negative after neoadjuvant chemotherapy. It is important to note that these factors were relatively rare and did not significantly impact the feasibility of liver transplantation in previous studies [11,18]. This observation implies that hepatoblastoma patients with an advanced local stage can be managed successfully with minimal impact on survival.
The CHIC-HS system serves as a risk stratification tool for hepatoblastoma patients [7,8]. In our study, incorporating AFP levels after the fourth cycle of neoadjuvant chemotherapy into the CHIC-HS system substantially enhanced its predictive power for 5-year EFS. A higher AFP level after neoadjuvant chemotherapy indicates the presence of a residual tumor [19], which might not be detectable through CT or MRI scans. This observation could explain why the additional inclusion of AFP levels after neoadjuvant chemotherapy to the CHIC-HS system significantly enhanced the prognosis prediction.
We analyzed inter-reader agreement between two radiologists for PRETEXT and POSTTEXT groups and annotation factors. The overall level of agreement was higher within the POSTTEXT system than the PRETEXT system. Among the PRETEXT factors, we observed good agreement in the order of M, F, and P, known as significant prognostic factors in previous studies [6,20]. The aggregation factor VPEFR, crucial within the context of the CHIC-HS system, exhibited relatively poor agreement. POSTTEXT factor M emerged as the sole significant prognostic factor among the POSTTEXT group and annotation factors. This factor exhibited substantial inter-reader agreement, suggesting its feasibility for clinical application.
There are several limitations in this study. First, a potential bias may exist due to its retrospective nature and the reliance on data from a single tertiary hospital cohort. However, given the paucity of hepatoblastoma cases, this study design was inevitable. Second, the study spanned an extended period, resulting in relatively shorter follow-up periods for patients in the later period. However, since most events generally occur within 3 years of diagnosis, a more extended follow-up period may not significantly alter the results.
In conclusion, the AFP level, the change ratio of tumor size and presence of metastasis after the fourth cycle of neoadjuvant chemotherapy (POSTTEXT M) emerged as significant predictors of EFS in children with hepatoblastomas. Including AFP level after the fourth cycle of neoadjuvant chemotherapy improved the predictive ability when combined with the CHIC-HS. This suggests that post-treatment evaluation might be focused on the presence of distant metastasis, tumor size change and AFP levels rather than thorough scrutiny of the extent of hepatoblastoma itself. Our results imply the potential for a new risk stratification model combining post-treatment factors in hepatoblastoma patients undergoing neoadjuvant chemotherapy. However, this model requires further external validation on a larger cohort in future studies.
XML Download