Hepatic adenoma regression after bariatric surgery: a case series and systematic review.

Materials and methods

Case series
This is a retrospective case series of adult patients from a single, quaternary academic medical center between March 2012 and January 2024. The study followed STROBE guidelines (Supplementary Table 1) and was approved by the Institutional Review Board [28]. Inclusion criteria included age > 16 years, patients who underwent either SG or RYGB, had evidence of HA prior to bariatric surgery, and had post-operative abdominal imaging at least six months after their index bariatric surgery. Patients who underwent hepatic resection, as well as patients without both pre- and post-operative imaging, were excluded.
Female patients who underwent bariatric surgery within the study’s timeframe were identified from the institution’s Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MBSAQIP) database. Medical records, including radiology and pathology reports, were screened for ICD codes and keywords for liver lesions (Supplementary Table 2). These records were then independently reviewed by the authors for evidence of HAs on pre-operative imaging. Only patients who had evidence of HA on pre-operative as well as post-operative imaging were included.

Systematic review
Given the lower-than-expected number of patients meeting inclusion criteria in our institutional series, a systematic review was also performed to identify publications reporting changes in HA size after bariatric surgery. PubMed, Embase, Cochrane Library, and Scopus databases were queried using the following search terms: liver cell adenoma, liver adenoma, adenoma, hepatic adenoma, hepatocellular adenoma, benign hepatoma, telangiectatic focal nodular hyperplasia; bariatric surgery, gastrectomy, gastric bypass, gastric bypass surgery, gastric sleeve, sleeve gastrectomy, weight loss surgery. Multiple combinations of search terms were used, and both free text and controlled vocabulary (MeSH and Emtree) were included. The initial search was completed on November 11, 2024.
All English language studies that examined patients who underwent primary bariatric surgery and had evidence of HA, including its size before and after surgery, were included. Studies investigating pharmaceutical and/or lifestyle interventions without surgical intervention were excluded, as were studies reporting patients who underwent hepatic resection only. Reviews, letters, commentaries, narratives, and study protocols were also excluded. Abstracts were allowed if there were no full-text duplicates. No restrictions were placed on the date of publication. Additional relevant studies were identified through manual searches of the literature.
Initial screening of eligible studies was performed through a review of titles and abstracts, followed by a secondary screening of the full texts. Each phase of screening was conducted by two independent reviewers (AKW, ICG), and disagreements were settled by consensus. The literature search and screening were completed using Covidence (Melbourne, Australia). The systematic review was performed according to PRISMA guidelines.

Data extraction
Data extraction from the included studies was undertaken by two authors (AKW, ICG). Variables extracted included publication year, country of origin, and study type. Patient data extracted included gender, age, sex, pre- and post-operative BMI, pre-and post-operative hepatic adenoma size, follow-up interval, type of bariatric surgery performed, diagnostic and follow-up imaging modality, and pathology reports as applicable. Hepatic adenoma sizes were determined by greatest dimensions seen on cross-sectional imaging or ultrasound unless intraoperative measurements were noted.

Quality appraisal
Each study was assessed for quality and presence of bias using the Joanna Briggs Institute (JBI) Critical Appraisal checklists by two independent reviewers (AKW, ICG), and disagreements were settled by consensus [29]. The JBI checklists for case reports and case series contain a series of “yes”, “no”, “unclear”, or “not applicable” questions to evaluate study design. Cutoffs at 70% and 50% of questions answered “yes” were used to determine high- and medium-quality studies, respectively. No studies were omitted based on quality assessment to ensure the entire body of literature was represented in this review.

Statistical analysis
Percentage of excess weight loss (% EWL) was calculated for each patient based on the following formula: . For these calculations, ideal BMI was set at 25 kg/m2. Scatter plots with Kendall-Theil-Sen estimators, given the bounded and non-parametric nature of this data set, were used to visualize the data. For non-parametric data, Kendall’s Tau-b was calculated as appropriate. Fractional regression with probit link was also conducted. Regression coefficients and effect sizes in the probit model are not easily interpreted. Therefore marginal effects representing the change in the size of hepatic adenomas using the delta method to calculate 95% confidence intervals were performed. All p values used in the analyses were two-tailed, and p values < 0.05 were considered significant. Data were analyzed by ICG from November 2024 to January 2025 using Stata 18 (Stata Corporation, College Station, TX).

Results

Institutional cohort
Among the 3,218 initial records retrieved, 827 unique patient charts were reviewed, ultimately resulting in five patients who underwent bariatric surgery and had evidence of HA on pre-operative imaging. Two records did not include post-operative cross-sectional imaging and were therefore excluded (Fig. 1). Patient characteristics of the institutional cohort are shown in Table 1.
Patient A, a 37-year-old female with pre-operative BMI of 46.92 kg/m2 underwent RYGB. She was noted to have three hepatic adenomas, each approximately one cm on MRI obtained due to right flank pain approximately one year prior to her bariatric surgery. The operation was uneventful; no obvious concerning changes to the hepatic surface were noted intraoperatively. Approximately 1.5 years after surgery, her BMI had decreased to 29.87 kg/m2. At that time, a computed tomography (CT) abdomen and pelvis (A/P) scan with contrast was obtained due to persistent nausea and emesis and showed no evidence of hepatic lesions.
Patient B, a 42-year-old female with a pre-operative BMI of 38.03 kg/m2, underwent RYGB. MRI obtained three years prior to her bariatric surgery showed a single HA measuring 1.8 cm × 1.4 cm in segment II, which was stable on CT immediately prior to her RYGB. The patient’s BMI decreased to 23.73 kg/m2 at 16 months post-operatively. Cross-sectional imaging obtained at the 16-month interval showed resolution of her HA.
Patient C, a 35-year-old female with a pre-operative BMI of 40.76 kg/m2, underwent sleeve gastrectomy. On pre-operative CT, a small 1.3 cm lesion was identified in segment IVb. MRI 18 months after SG showed a 1.1 cm HA in the same area as well as a small hepatic hemangioma in segment VII. The patient’s BMI at the time of her most recent MRI was 36.2 kg/m2.

Systematic review cohort
Among the 1,840 publications retrieved in the initial query, including three from searching the reference list of relevant studies, 612 were identified as duplicates. Titles and abstracts were screened for the remaining 1,228 studies. Ultimately, after a full-text review of 10 articles, results were extracted from five studies that met inclusion criteria (Fig. 2). The literature search identified four case reports and one case series. All studies were from different countries, including the United Kingdom, Brazil, Qatar, the Netherlands, and France. Of the five studies included in the review, all had reporting rates of > 75%, with 4/5 studies having 100% complete reporting rates per JBI guidelines (Supplementary Table 3). From these studies, seven patients were identified, all of whom were female (Table 2). Four patients underwent RYGB, two underwent SG, and one underwent gastric banding.
Gevers et al. (2018) described a case series of three obese patients with known HA who underwent RYGB. In all cases, patients lost significant excess body weight (31–48%) post-operatively and showed decreased size, or complete resolution, of their HA based on follow-up cross-sectional imaging [25].
Khaoudy et al. (2018) described the case of a 27-year-old female with a BMI of 46.5 kg/m2 and biopsy-proven HA (inflammatory subtype) noted to be 11 cm × 8.1 cm. This lesion was located near the middle hepatic vein, and resection would have required a right or central hepatectomy. This patient initially trialed medical weight loss with a decrease of 9.1% of excess weight, corresponding to a slight reduction in HA size to 10 cm × 7.7 cm. Given the persistently large size of the hepatic adenoma and elevated body weight, the patient subsequently underwent sleeve gastrectomy to provide additional weight loss and tumor control. Post-operatively, the patient lost 63.6% of excess weight with a corresponding decrease in adenoma size to 4.3 cm × 3.7 cm on follow-up imaging [26].
Bonanomi et al. (2008) described the case of a 34-year-old female with a BMI of 40.4 kg/m2 who underwent gastric banding [30]. Intraoperatively, the patient was noted to have a four cm nodule in segment III. An intraoperative biopsy of this lesion was performed, but the results were inconclusive, as histologic features could not differentiate between HA and focal nodular hyperplasia. The patient’s initial post-operative MRI suggested that the lesion identified intraoperatively was likely an HA. The patient discontinued oral contraceptives and proceeded to lose 24 kg in the subsequent six months for a 57.4% decrease in excess weight to 31.6 kg/m2. Follow-up MRI six months after gastric banding did not show any decrease in the size of the HA. The patient subsequently underwent laparoscopic hepatic resection, with final pathology of the surgical specimen consistent with HA.
Dantas et al. (2021) described the case of a 47-year-old female with a pre-bariatric surgery BMI of 41.4 kg/m2 who was noted to have a large, eight cm hepatic adenoma between segments VII and VIII found on routine abdominal ultrasound [31]. The patient subsequently underwent an uneventful RYGB. The patient initially maintained a 70.3% reduction in excess weight at one year with a concomitant resolution of the HA. While the patient regained weight between years one and five post-operatively, with a decrease to 66.5% reduction in excess weight at year five post-operatively, MRI obtained at that time did not have evidence of HA.
Dauleh et al. (2022) described the case of a 17-year-old female with Prader-Willi syndrome, a BMI of 64.4 kg/m2 who presented with liver adenomatosis (≥ 10 HAs), the largest measuring 1.3 cm in diameter [32]. Despite numerous attempts at non-surgical weight loss including diet, exercise, and a trial of glucagon-like peptide 1 receptor (GLP-1) agonist, the patient experienced persistent weight gain. She subsequently underwent a sleeve gastrectomy with a resulting 7.8 kg weight loss two months post-operatively without change in the sizes of her HAs.

Combined cohort
The combined institutional and systematic review cohort consisted of 10 female patients with an average age of 34.8 years, ranging from 17 to 47 years old at the time of bariatric surgery. The average (SD) pre-operative BMI was 44.48 (7.42) kg/m2, and the average post-operative BMI was 32.19 (10.08) kg/m2, corresponding to an average (SD) reduction in BMI after surgery of 28.41 (12.60) % and mean (SD) %EWL of 68.18 (9.49) %. The mean size of the largest HA was 4.49 cm pre-operatively and 1.55 cm post-operatively (mean (SD) change in size -56.46% (43.68)).
To visualize the association between the change in BMI and the percent change in HA size after bariatric surgery, a scatter plot was created, and a Kendall-Thiel-Sen estimator was fitted to the data (Fig. 3a). A similar plot was created to visualize %EWL vs percent change in HA size (Fig. 3b). Weight loss, as defined as percent decrease in BMI, was not statistically significantly associated with percent reduction in HA size (Kendall’s Tau-b: 0.4969; p = 0.0735), although it trended toward significance. However, changes in %EWL were statistically significantly associated with percent reduction in HA size (Kendall’s Tau-b: 0.5528; p = 0.0471). Fractional probit regression analysis was utilized to analyze variables predictive of changes in HA size post-operatively. In this model, %EWL is associated with decreased HA size (p = 0.004; 95% CI -7.46, -1.45). Average marginal effects were then calculated, suggesting that each additional percent increase in %EWL is associated with 0.76% HA size reduction (95% CI: -1.18, -0.34; p < 0.001) when holding age constant at the mean for the population (Table 3).

Discussion
Despite decreased use of oral estrogen-containing contraceptives, the prevalence of HAs has not declined correspondingly, underscoring the role of obesity in HA formation [4, 33]. Given the established association between HA formation and obesity, interventions aimed at weight loss could represent a rational initial therapeutic approach for many patients with HAs. Nevertheless, this systematic review of the literature identified a relative scarcity of published series on the role of weight-loss surgery among patients with HAs. Combined with data from our own institutional series, we noted that the percentage of excess weight loss (%EWL) following bariatric surgery was significantly associated with change in HA size. Notably, 40% of patients experienced complete regression of their HA. Given the additional benefits of weight loss from bariatric surgery and the significant morbidity associated with liver resection, bariatric surgery could be a promising strategy for obese patients with HA.
While benign, HAs maintain the potential for both malignant transformation and hemorrhage. For this reason, surgical resection has traditionally been offered to patients who meet specific clinical criteria, such as the presence of symptomatic HA or lesions that are greater than five cm in diameter [10]. Given the morbidity associated with hepatic surgery even in generally young, healthy women, there has been considerable interest recently in nonoperative management. For example, current guidelines now typically recommend cessation of OCPs as the preferred initial management even for HAs that meet resection criteria [34]. However, for obesity-related HA, evidence supporting the role of weight loss as a management strategy is less robust [27, 35]. In a large institutional series of non-surgical management of HA (including OCP cessation, medical weight loss through calorie reduction as well as exercise, and arterial embolization) in patients with a BMI ≥ 35 kg/m2, only two patients experienced regression of their HAs with diet and exercise alone. Both patients had significant weight loss (31% reduction in BMI) [27]. Along with the results of our study, this finding suggests that dramatic changes in %EWL are necessary to induce regression of HA. Indeed, the few patients in our study who did not achieve a reduction in their HA size, experienced minimal weight loss during the follow-up period.
The exact mechanism of HA regression after bariatric surgery is unknown. However, the reduction of circulating androgens and improved metabolic profile after bariatric surgery likely contribute. Bariatric surgery is associated with decreased hepatic steatosis, thought to be secondary to reduced hepatic inflammation, improved insulin sensitivity, and improved gut microbiota [36–40]. The reduction of persistent inflammation of hepatocytes, improved hepatic metabolism, and decrease in systemic inflammation through an improved hormonal milieu likely promotes regression of HAs. Over the past several years, there has been an inundation of new, effective weight-loss medications on the market. These glucagon-like peptide-1 (GLP-1) receptor agonists function by stimulating endogenous insulin release, inhibiting gastric emptying, and inhibiting endogenous glucagon release [41]. Given the relatively recent approval of these drugs for weight loss, no studies to date have investigated their impact on HAs. However, existing evidence from animal models and retrospective studies in patients with diabetes demonstrates a decrease in transaminitis and hepatic steatosis with use of GLP-1 receptor agonists [42, 43]. This suggests that GLP-1 receptor agonists may be beneficial for treating HAs through multiple mechanisms. By promoting weight loss, these agents may reduce levels of circulating androgens, which are implicated in HA pathogenesis. Additionally, obesity can lead to a constitutively pro-inflammatory state characterized by elevated IL-6 production. High levels of IL-6, in the setting of chronic inflammation, can contribute to HA formation, specifically the inflammatory subtype of HA [4, 44]. The reduction of hepatic steatosis seen with GLP-1 receptor agonist use, potentially through a reduction in hepatocyte sensitivity to circulating pro-inflammatory cytokines or reduced levels of these cytokines, may support HA regression, although more research is needed to elucidate these pathways fully.
Despite the large volume of young obese women undergoing primary bariatric surgery at our institution, a large quaternary medical center, the prevalence of HAs was relatively low during the study period (i.e. only 0.61% of female patients). Within the institutional cohort, only five patients were noted to have HA prior to bariatric surgery, and only three of those patients had appropriate post-operative imaging within the follow-up period. As patients undergoing bariatric surgery are in themselves a selected group, further research is needed to estimate the incidence of HA among obese persons overall.
While our study suggests a benefit of bariatric surgery and weight loss in general for HA regression, the routine use of bariatric surgery for the management of obesity-associated HAs may be challenging. Previous research has highlighted the barriers to bariatric surgery, including lower socioeconomic status, non-English language as the primary language, rural residence, uninsured or underinsurance status, and older age [45–49]. Strategies will be needed to overcome these barriers for bariatric surgery to be a feasible option for managing HAs.
A strength of our study, despite its small sample size, is the systematic review and inclusion of all published reports on this topic. Along with our own case series, we report the largest combined cohort of patients with HA undergoing bariatric surgery. Still, several limitations should be acknowledged. Due to the study design, follow-up time, imaging modality, type of bariatric surgery, indications for treatment, and more were not standardized across the cohort. In addition, given the nonrandomized nature of the study, there may be other confounding factors, and a causative relationship between bariatric surgery and HA regression cannot be definitely concluded. Finally, information on HA subtype was not available, and therefore the association between weight loss and HA regression in different subtypes requires further investigation.
In conclusion, using a combined institutional series and systematic review of existing literature, we report the largest series of patients with HA undergoing bariatric surgery. A strong association between %EWL after bariatric surgery and HA regression was observed, suggesting that bariatric surgery could be considered as an initial management option for patients with obesity-related HA. Future prospective trials investigating the role of intentional weight loss for hepatic adenoma regression are warranted.

Supplementary Information
Below is the link to the electronic supplementary material.