The value of liver resection for focal nodular hyperplasia: resection yes or no?

Focal nodular hyperplasia (FNH) is the second most common benign lesion of the liver most frequently occurring in healthy, young and middle-aged women [1‐3]. As most FNH are asymptomatic and rarely grow, these lesions are often discovered incidentally by routine abdominal ultrasound [3‐5]. Although malignancy can safely be ruled out in most cases based on imaging and surveillance of these patients, some cases represent diagnostic challenge even in these days [3‐7]. In this context, many different single-center algorithms for the diagnostic and therapeutical management of asymptomatic patients with incidental benign liver tumors have been initiated within the past years, and a consensus paper has recently been published to manage these findings on computed tomography (CT) in asymptomatic patients [4, 8, 9].

Diagnostic tools commonly used over the past years include ultrasound, nuclear medical procedures (hepatobiliary scintigraphy with ⁹⁹Tc) and cross-sectional imaging, e.g. CT. Contrast-enhanced magnetic resonance imaging (MRI) scanning has been shown to be the most sensitive modality for the characterization of FNH and is increasingly applied [10‐15]. However, characterization might still be challenging, e.g. for single lesions being too small to characterize (TSTC lesions) or in patients with a history of cancer [13‐18].

Furthermore, the differential diagnosis can be difficult in cases of atypical appearance (e.g. missing central scar) or in elderly patients with underlying chronic liver disease at risk for hepatocellular carcinoma (HCC) [16‐18]. Invasive diagnostics, e.g. percutaneous biopsy of hypervascular lesions might be associated with bleeding and tumor spread in case of malignancy. Theses techniques may also show unclear or misleading results [2‐4, 19‐21].

The management of benign liver tumors has changed over time and is still evolving. Considering the indolent natural history of FNH with a low risk for complications and no malignant potential, patients with asymptomatic FNH should indeed be treated conservatively [1‐6, 11, 12, 19, 20, 22]. A causal relationship between hormonal contraception and growth of FNH has broadly been discussed controversially by many authors, whereas a minority of physicians might still favor and recommend at first to discontinue hormonal contraceptives [22‐24].

Surgical approaches should be considered in case of tumor enlargement (after discontinued hormonal contraception), as these patients are at increased risk for intrahepatic complications, e.g. bile duct compression with resulting cholestasis and for rupture or/and acute bleeding [2‐6, 10, 20, 22, 25, 26]. Liver resection might also be indicated if FNH presents with atypical features and malignancy cannot be ruled out [22‐26]. Some patients indeed report on severe lesion-associated symptoms, e.g. abdominal pain or obstruction of large vessels or intrahepatic bile ducts and might further benefit from surgical approaches. However, many patients suffer from unspecific abdominal discomforts and it remains uncertain whether these patients really benefit from liver resections [10, 20, 22, 26‐29]. Unfortunately, there are only limited data evaluating the benefit-risk balance after liver resections for benign liver lesions [30, 31]. Previous studies investigating QoL after liver resection mostly included patients with malignant tumors [32, 33]. However, to date there are only few studies available investigating QoL improvements after liver resection for FNH [22].

We here report on 100 consecutive patients undergoing liver resection for FNH within the past 20 years at a single center. The retrospective analysis summarizes indications for surgical approaches and the outcome of these patients, especially with regard to their QoL after liver resection. Liver resection for FNH represents a safe approach, being associated with low morbidity and no mortality in our series. Especially symptomatic FNH patients might benefit from liver resection, as shown by significant improvements of their QoL.

This study represents one of the largest retrospective analyses, examining patients undergoing liver resection for FNH to date. Despite remarkable advantages in diagnostic modalities within the last years, there remains a considerable proportion of misleading preoperative diagnosis, especially in patients with atypical forms of FNH and history of cancer. Reasons for surgery included abdominal discomfort, uncertainty of diagnosis/malignancy and tumor enlargement/jaundice. Our results clearly indicate that liver resection can be performed safely with major complications evident in only 8 % and no mortality. Surgery for FNH is associated with a high patient satisfaction and in our study improved QoL, especially in symptomatic patients.

There is an emerging interest in managing incidental findings in the liver and pancreas [3‐5, 7‐9, 37]. Among cross-sectional imaging, MRI is considered the most reliable imaging method for classifying incidental liver lesions, especially after recent improvements, e.g. the introduction of the new T1-positive liver specific contrast agent, gadoxetic acid (Gd-EOB-DTPA, Primovist or Evosit, Bayer Schering Pharma, Germany) [13‐18]. However, even after combining different non-invasive imaging modalities, e.g. CT and MRI, and invasive percutaneous fine-needle biopsy, diagnostic uncertainty remains in up to 40 %. In accordance with the current literature, we noted incorrect/uncertain findings on CT and MRI in 58 and 41 %, respectively, with 40 % remaining incorrect/uncertain after combined both diagnostic approaches [10, 11, 20]. When the CT was incorrect/uncertain, running an additional MRI did not help establish a more correct preoperative diagnosis in our study. There were no statistical significant improvements with regard to the percentage of correct diagnosis over the investigated time periods. Our data show a non significant decrease in surgical procedures between 1992 and 2012 in patients with histologically proven diagnosis of FNH. The apparent decrease of carried out surgical procedures does not necessarily represent a decrease in incidence of FNH, but only a decrease of patients who underwent surgical procedures for FNH. The reason for the decrease of surgical interventions in FNH patients lies, in our opinion, in the more accurate and reliable diagnostic modalities. Of note, percentage of incorrect/uncertain diagnosis by CT and MRI is likely invalid high, as the patients’ collective does not represent all patients with FNH, but only patients undergoing surgery.

Percutaneous fine-needle biopsy represents another diagnostic possibility in the workflow for uncertain liver lesions. However, its significance is discussed controversially, as it is associated with an increased bleeding risk in hypervascularized lesions and a risk of peritoneal seeding in case of malignancy [2‐7, 19‐21]. Further, tumor biopsy is associated with a low diagnostic sensitivity, with only 30–45 % of all biopsy being consistent with the histology of surgical preparations [2, 3, 21]. This could be confirmed in our patients’ cohort, where only in 10 of 21 fine-needle biopsies (47 %) were confirmed by histology postoperatively. Apart from operative management strategies of symptomatic FNH, percutaneous radiological modalities have to be considered and include arterial embolization and radiofrequency ablation (RFA), which have been published in a few case series [10, 38‐40]. However, up to date no randomized controlled trials exist comparing the outcome of surgical resection with percutaneous techniques. Of interest, the major limitation of these interventional techniques is the lack of post-procedural histology. Therefore, these techniques should only be applied if a definitive diagnosis of FNH could be secured by preoperative imaging and biopsy-derived diagnosis [10].

Most incidentalomas are likely benign without or little clinical significance. Committees have recently been formed to manage incidental findings on CT and consensus guidelines try to classify patients groups of high risk with hepatic and pancreatic lesions [8, 9, 28, 37]. Factors like age (>40 years), history of malignancy, hepatic malignant risk factors as well as hepatic dysfunction with or without symptoms help to define patients of considerably high risk for liver malignancies [8, 9, 37]. In accordance, surgery was performed in our cohort because of uncertainty of diagnosis/malignancy in 41 patients, cancer history in 18 patients and tumor enlargement/jaundice with a rate of growth of >0.5 cm/year or 2–3 cm in comparison with initial size in 13 patients.

Certainly, special guidelines for benign liver lesions are needed to balance cost-intensive long-term follow ups by MRI against the risk of unnecessary operations associated with perioperative complications. We need to take in account not only improvements in radiological findings; due to ongoing improvements in liver surgery with decreased perioperative complications even major hepatic resections can be safely performed in experienced hepatobiliary centers [19, 22, 29, 30, 41‐43]. Based on our results with low perioperative major complications (8 %) and no mortality, this might influence the process making a decision in these patients. In contrast to asymptomatic patients diagnosed with incidentalomas of the liver, 46 patients underwent liver surgery because of abdominal discomfort with nonspecific symptoms (nausea, fatigue, decreased appetite, etc.) or pain. Although the evaluated symptoms represent typical complaints of patients with benign hepatic tumors, they are admittedly not specific for FNH. However, the symptoms listed in our questionnaire were chosen in support of similar studies evaluating QoL in patients with surgery for benign hepatic tumors and therefore seem suited to evaluate QoL postoperatively [30, 31]. Abdominal pain as indication for surgery can be challenging, as it is difficult to provide assurances that symptoms will improve after surgery [22, 30]. Therefore, liver resection should be considered only after exclusion of other causes for abdominal symptoms and if a relief of symptoms is expected after surgery. Only a few reports on QoL improvements after liver resection of benign liver tumors exist, with no sufficient data for FNH available [22, 30, 31, 41, 43].

Beside a decreased complication rate after liver surgery, long-term outcome represents an important factor for evaluating patients for surgery. Our study provides first evidence that QoL might improve in patients with symptomatic FNH after liver resection.

The mean patient self-reported pain levels had significantly decreased over time. In addition to pain and other preoperative symptoms, several further QOL domains including physical health, mental/social/emotional health such as general health were evaluated. We noted improvements in a wide range of QOL domains with 20 % of our patients reporting a significant improvement of general health after liver resection. Furthermore, significant improvements were noted not only in physical activities, but also in social and mental health. These improvements resulted in greater work productivity, increased energy level and better social functioning. These findings together with the fact that 88 % of patients were generally satisfied and would undergo liver surgery again seem to prove a positive effect of liver surgery on daily life of the patients.

Regarding the different types of surgical approaches, e.g. laparascopic vs. open surgery, there were no statistically significant differences with regard to reported pain levels over time. This is in contrast to previous studies by Kneurtz et al., who noted markedly better pain scores after laparascopic surgery 6 and 12 months after liver resection [31]. This controversy might be due to the fact that only patients scheduled for minor resections were evaluated for laparoscopic approaches in our cohort. In accordance with similar studies, patients undergoing laparoscopic liver resection reported a postoperative improvement of QoL of 2.3-fold compared to patients undergoing open surgery [31]. Even though an interpretation of the QoL data is limited by its retrospective character, these results give us a strong impression of patient-related outcome after resection of benign liver tumors and illustrate the benefits of surgery from patient’s point of view.

The current study has several limitations that should be considered. First, data were collected and analyzed retrospectively with a considerable time period between surgical procedure and patients’ interview with a response rate of approximately 60 %. In this context, the collection of data may be influenced by recall bias. For example, a possible concern might be that patients with better operative outcomes (e.g.: reduction of postoperative pain score, no presence of malignant tumor in final postoperative histology, no operative complications) answered the QoL survey more often than patients with inferior postoperative outcome. This bias could lead to an overestimation of the postoperative improvement of QoL in our study. In our study, we found an overall complication rate of 19 % concerning all conducted surgeries in benign hepatic tumors. The patients who answered the QoL questionnaire had an overall complication rate of 16 %, comparably to the overall complication rate. Therefore, if one may consider the complication rate as parameter for negative outcome or negative association of the surgical intervention by the patient, a similar complication rate might contradict the suggested recall bias. Another limitation might be that patients with the pre-existing diagnosis of a malignant tumor might feel more worried by the presence of a malignant tumor in the liver and postoperative benign diagnosis may have brought these patients more relief compared to patients without a cancer history. Patients with a subjective uncertainty of malignancy had a non significant increase in QoL after hepatic surgery for FNH. However, all other patients with preoperative proven diagnosis of a benign tumor nature also showed an insignificant increase of QoL. Therefore, we suggest that the histologically proven certainty of benign tumor nature may play an important role in QoL for patients, who are worried about the potential malignancy of the tumor; however, patients without worries about tumor malignancy also profited from the operation in our study.

This study demonstrates that liver resection for benign liver tumors can be performed safely in specialized hepatobiliary centers. Despite improvements in diagnostic modalities, there remain liver lesions that cannot be specified reliably by imaging. In case of uncertain diagnosis, especially in patients with a history of malignancy or suspected hepatocellular adenoma, marked tumor enlargement and/or jaundice, surgical resection should be considered. In case of symptomatic liver lesions, surgical resection should only be indicated in patients with tumor-specific symptoms. Regarding our results, surgery for FNH is associated with marked improvements in patient-reported pain scores as well as other QoL domains. Patients with significant preoperative symptoms show the most benefit from surgical intervention.