Factors Influencing Overall Survival for Patients with Fibrolamellar Hepatocellular Carcinoma: Analysis of the Surveillance, Epidemiology, and End Results Database.

BACKGROUND: The study aimed to develop a nomogram to predict overall survival (OS) for patients with FLC using a national database.
METHODS: The Surveillance, Epidemiology, and End Results database of the National Cancer Institute was reviewed to identify FLC cases with histological confirmation between 2004 and 2014. Cox proportional hazards models were used to identify factors associated with OS. The validation of the nomogram was performed using concordance index (C-index) and calibration curves.
RESULTS: Out of 170 cases with complete follow-up, 87 received surgery/ablation and 12 received transplantation with significantly higher OS than chemotherapy alone while transplantation combined with chemotherapy showed better survival than solely transplantation. The combination of surgery and chemotherapy showed worse OS than surgery alone. Survival was negatively influenced by T4 stadium (HR = 5.91), while young age and surgery were positive predictive factors. There was no influence of gender, ethnicity or nodal status on survival. The rate of AFP positivity was comparable with and without the presence of distal metastases.
CONCLUSION: FLC survival is greatly dependent upon appropriate surgical management irrespective of tumor stadium.

Background

Fibrolamellar hepatocellular carcinoma (FLC) is a rare type of primary liver tumor, mostly presenting in younger patients without underlying liver disease and cirrhosis.1–3 Due to its rarity, it has initially been classified as hepatocellular carcinoma (HCC), but the affected population and the underlying factors differ. National incidence also presents with marked variability, with the previously reported estimated age-adjusted incidence rate of less than 1 per 100,000 in the United States but with noticeably higher incidence in other countries such as Mexico (5.8% of liver tumors).4 The most commonly reported age span for the diagnosis of FLC is under 40 years of age.5 Diagnosis of FLC is mainly dependent on imaging since clinical symptoms are non-specific (abdominal pain is the most common symptom) and common biomarkers for liver disease are often negative. FLC presents most often as an enlarged single lesion with a centrally located scar in two-thirds of the cases. This makes the differentiation with other liver tumors (ie focal nodular hyperplasia) more difficult.6 Other entities which resemble FLC are adenomas or metastases of neuroendocrine tumors.7 The diagnosis is based on cross-imaging studies of CT and MRI, but a liver biopsy is recommended also for tumors that are deemed unresectable since there is a substantial risk of misdiagnosis.8–11

Timely resection of FLC is the main aspect of treatment; most patients are in need of extended liver resection since nodal and vascular infiltration have been reported as high as 60% and 43%, respectively.12–14 The option of liver transplantation has also been used for selected cases with good prognosis; however, there is little priority for patients with non-HCC liver tumors on the waiting list.15,16 A major prognosis burden is the fact that FLC presents high recurrence rates.17,18 Taking into account all these factors there is a need for better prognostic tools for FLC. Our goal was to assess the survival trends of FLC and construct a nomogram for overall survival (OS).

Methods

Data Source

A registry-based retrospective cohort study was performed using data from the SEER database for the years 2004–2014. SEER is supported by the Surveillance Research Program (SRP) in NCI’s Division of Cancer Control and Population Sciences (DCCPS). SEER Program provides information on comprehensive demographic and cancer statistics covering 28% of the US population.19 The characteristics and representativeness of this database have been discussed previously.20,21

Study Population

Between January 1, 2004, and December 31, 2014, 199 patients were diagnosed with FLC. This calendar year interval was chosen to provide a sufficient information on tumor characteristics and chemotherapy use. Cases were identified using a specific histologic code (International Classification of Diseases for Oncology [ICD-O] code 8171). Patients who had no confirmed pathology diagnosis (N=17), diagnosed at autopsy or by death certificate (N=1) and had another malignant primary tumor (N=11) were excluded. All cases were restricted to first primary FLC. A total of 170 patients were selected for further analysis.

The following variables were retrieved from the SEER database: sex, age (categorized as age <18, 18–39, 40–64, and >65 years), ethnicity (Caucasian, African-American, other (American Indian/Alaska native, Asian/Pacific islander)), year of diagnosis, insurance status, tumor multiplicity and size, tumor stage, AFP status, chemotherapy use and surgical management. Because all patient information in the SEER database is de-identified, this study was exempted from review by the Institutional Review Board.

Statistical Analysis

We conducted a univariate analysis for surgical approach, chemotherapy, and the combination of these approaches as a categorical outcome with a list of covariates. Univariate analysis was undertaken using validated dataset macros.22 The p-value was calculated by ANOVA for numerical covariates and chi-square test or Fisher’s exact for categorical covariates, where appropriate. The primary end point in this study was overall survival. OS was defined as the time from diagnosis to death, date of last follow-up. Kaplan–Meier methods and the Log rank test were utilized for the survival analysis. A Cox proportional hazards model was applied for multivariate survival analysis. Candidate variables with a p value <0.01 on univariate analysis for survival were subjected to multivariable Cox regression analysis. In addition, to adjust for the potential confounding effect due to sex, the multivariable model was fit where sex was included as a covariate.

The performance of the nomogram was measured by concordance index (C-index) and assessed by comparing nomogram-predicted versus observed Kaplan–Meier estimates of survival probability rates. Bootstraps with 1000 resample were used for internal validation purposes. Bootstrapping was used to obtain a relatively unbiased estimate. Statistical analyses were carried out with SAS version 9.4 for Windows (SAS Institute Inc., Cary, NC, USA) and R version 3.5.3 (). All tests of statistical significance were 2-sided, and statistical significance was established at P < 0.05.

Results

Demographics

Young adults aged under 40 years had the highest incidence (70%) with the majority of patients being Caucasian males. The 3-year incidence of the disease showed a marked reduction of 12.3% over the three periods of study. At baseline, most tumors had surpassed the size cut-off of 5cm (68.2%) while most patients had single lesions. Vascular infiltration was present at 39.4% of tumors and most tumors did not present with positive AFP values at baseline.

More than half of the patients received surgical treatment or transplantation, but we observed a trend towards less surgical treatment over the study period. Between 2004–2007 and 2008–2011, less patients received surgery or transplantation, but this was not statistically significant (chi-squared test p=0.104). Half of the patients received chemotherapy in this study cohort. The detailed demographics of the study sample are shown in Table 1.

Table 1

Descriptive Statistics for All Variables

Variable	Level	N (%) = 170
Age at diagnosis	Mean ± SD	33.28 ± 20.38
Age category	<18	42 (24.7)
	18–39	77 (45.3)
	40–64	32 (18.8)
	>65	19 (11.2)
Gender	Male	106 (62.4)
	Female	64 (37.6)
Year of diagnosis	2004–2007	67 (39.4)
	2008–2011	57 (33.5)
	2012–2014	46 (27.1)
Insurance Status	Uninsured	3 (1.8)
	Insured	120 (70.6)
	Unknown	47 (27.6)
Ethnicity	Caucasian	131 (77.1)
	African-American	21 (12.4)
	Other	18 (10.6)
AFP	Negative	78 (45.9)
	Positive	53 (31.2)
	Unknown	39 (22.9)
Tumor size	≤5 cm	29 (17.1)
	>5 cm	116 (68.2)
	Unknown	25 (14.7)
Number of lesions	Single lesion	102 (60.0)
	Multiple lesions	46 (27.1)
	Unknown	22 (12.9)
Distant metastasis	No	113 (66.5)
	Yes	51 (30.0)
	Unknown	6 (3.5)
Differentiation grade	G1	15 (8.8)
	G2	39 (22.9)
	G3	17 (10.0)
	Unknown	99 (58.2)
Vascular invasion	No	82 (48.2)
	Yes	67 (39.4)
	Unknown	21 (12.4)
Tumor stage	T1	71 (41.8)
	T2	22 (12.9)
	T3	47 (27.6)
	T4	14 (8.2)
	Tx	16 (9.4)
Node stage	N0	107 (62.9)
	N1	46 (27.1)
	Nx	17 (10.0)
Chemotherapy	No	85 (50.0)
	Yes	85 (50.0)
Surgery	No surgery or ablation	71 (41.8)
	Liver resection or ablation	87 (51.2)
	Liver transplantation	12 (7.1)
Survival (months)	Mean ± SD	33.82 ± 33.61

Note: All data presented as N (Col %) or Mean ± SD.

Univariate Analysis

We performed a univariate analysis of covariates regarding surgical approaches in order to infer which factors affect surgical treatment in FLC. Chemotherapy was added as both a variable and a factor in separate analyses in order to ascertain the effect of chemotherapy on whether surgical treatment is chosen. In the analysis of surgery, the majority of patients presented with a single lesion (70.1%). Apart from tumor multiplicity, the tumor nodal status and size were of significance; however, most patients in the surgical treatment group presented with lesions larger than 5cm (75.9% vs 19.5% with ≤5cm) (Table 2).

Table 2

Univariate Association with Study Cohort and Surgery

			Surgery
Covariate	Level	No Surgery N=71	Liver Resection/Ablation N=87	Transplantation N=12	P-value*
Age category	<18	13 (18.31)	25 (28.74)	4 (33.33)	0.128
	18–39	28 (39.44)	43 (49.43)	6 (50)
	40–64	18 (25.35)	12 (13.79)	2 (16.67)
	>65	12 (16.9)	7 (8.05)	0 (0)
Gender	Male	42 (59.15)	56 (64.37)	8 (66.67)	0.758
	Female	29 (40.85)	31 (35.63)	4 (33.33)
Year of diagnosis	2004–2007	23 (32.39)	40 (45.98)	4 (33.33)	0.120
	2008–2011	30 (42.25)	25 (28.74)	2 (16.67)
	2012–2014	18 (25.35)	22 (25.29)	6 (50)
Insurance status	Uninsured	2 (2.82)	1 (1.15)	0 (0)	0.451
	Insured	53 (74.65)	57 (65.52)	10 (83.33)
	Unknown	16 (22.54)	29 (33.33)	2 (16.67)
Ethnicity	Caucasian	51 (71.83)	69 (79.31)	11 (91.67)	0.625
	African-American	11 (15.49)	9 (10.34)	1 (8.33)
	Other	9 (12.68)	9 (10.34)	0 (0)
AFP	Negative	28 (39.44)	44 (50.57)	6 (50)	0.030
	Positive	30 (42.25)	18 (20.69)	5 (41.67)
	Other	13 (18.31)	25 (28.74)	1 (8.33)
Tumor size	≤5 cm	11 (15.49)	17 (19.54)	1 (8.33)	<0.001
	>5 cm	40 (56.34)	66 (75.86)	10 (83.33)
	Unknown	20 (28.17)	4 (4.6)	1 (8.33)
Number of lesions	Single lesion	35 (49.3)	61 (70.11)	6 (50)	0.022
	Multiple lesions	23 (32.39)	17 (19.54)	6 (50)
	Unknown	13 (18.31)	9 (10.34)	0 (0)
Differentiation grade	G1	4 (5.63)	10 (11.49)	1 (8.33)	0.011
	G2	8 (11.27)	26 (29.89)	5 (41.67)
	G3	7 (9.86)	10 (11.49)	0 (0)
	Unknown	52 (73.24)	41 (47.13)	6 (50)
Vascular invasion	No	34 (47.89)	45 (51.72)	3 (25)	0.096
	Yes	25 (35.21)	33 (37.93)	9 (75)
	Unknown	12 (16.9)	9 (10.34)	0 (0)
Tumor stage	T1	27 (38.03)	41 (47.13)	3 (25)	<0.001
	T2	4 (5.63)	18 (20.69)	0 (0)
	T3	21 (29.58)	18 (20.69)	8 (66.67)
	T4	6 (8.45)	8 (9.2)	0 (0)
	Tx	13 (18.31)	2 (2.3)	1 (8.33)
Node stage	N0	41 (57.75)	58 (66.67)	8 (66.67)	0.013
	N1	16 (22.54)	26 (29.89)	4 (33.33)
	Nx	14 (19.72)	3 (3.45)	0 (0)
Distant metastasis	No	33 (46.48)	70 (80.46)	10 (83.33)	<0.001
	Yes	33 (46.48)	16 (18.39)	2 (16.67)
	Unknown	5 (7.04)	1 (1.15)	0 (0)
Chemotherapy	No	29 (40.85)	53 (60.92)	3 (25)	0.009
	Yes	42 (59.15)	34 (39.08)	9 (75)
Age at diagnosis		39.21 ± 22.7122.71	29.26 ±17.75	27.25 ±15.6	0.005
Survival months		14.46 ±21.83	50.22 ±34.43	29.42 ±22.16	<0.001

Notes: *The p-value is calculated by ANOVA for numerical covariates; and chi-square test or Fisher’s exact for categorical covariates, where appropriate. P-values are printed in bold when under the 0.05 threshold.

Patients with distant metastases were treated more often with chemotherapy, although the database findings showed that 16 patients underwent resection and 2 underwent transplantation while distant metastasis was present. In the univariate analysis of chemotherapy, only age, N1, M1 and T were significantly associated with this treatment(Table 3).

Table 3

Univariate Association with Study Cohort and Chemotherapy

		Chemotherapy
Covariate	Level	No N=85	Yes N=85	Parametric P-value*
Age category	<18	13 (15.29)	29 (34.12)	0.023
	18–39	41 (48.24)	36 (42.35)
	40–64	18 (21.18)	14 (16.47)
	>65	13 (15.29)	6 (7.06)
Gender	Male	55 (64.71)	51 (60)	0.527
	Female	30 (35.29)	34 (40)
Year of diagnosis	2004–2007	33 (38.82)	34 (40)	0.322
	2008–2011	25 (29.41)	32 (37.65)
	2012–2014	27 (31.76)	19 (22.35)
Insurance status	Uninsured	2 (2.35)	1 (1.18)	0.128
	Uninsured	54 (63.53)	66 (77.65)
	Unknown	29 (34.12)	18 (21.18)
Ethnicity	Caucasian	68 (80)	63 (74.12)	0.124
	African-American	12 (14.12)	9 (10.59)
	Other	5 (5.88)	13 (15.29)
AFP	Negative	41 (48.24)	37 (43.53)	0.507
	Positive	23 (27.06)	30 (35.29)
	Other	21 (24.71)	18 (21.18)
Tumor size	≤5 cm	19 (22.35)	10 (11.76)	0.158
	>5 cm	53 (62.35)	63 (74.12)
	Unknown	13 (15.29)	12 (14.12)
Number of lesions	Single lesion	58 (68.24)	44 (51.76)	0.073
	Multiple lesions	17 (20)	29 (34.12)
	Unknown	10 (11.76)	12 (14.12)
Differentiation grade	G1	5 (5.88)	10 (11.76)	0.145
	G2	25 (29.41)	14 (16.47)
	G3	7 (8.24)	10 (11.76)
	Unknown	48 (56.47)	51 (60)
Vascular invasion	No	47 (55.29)	35 (41.18)	0.183
	Yes	29 (34.12)	38 (44.71)
	Unknown	9 (10.59)	12 (14.12)
Tumor stage	T1	42 (49.41)	29 (34.12)	0.041
	T2	14 (16.47)	8 (9.41)
	T3	16 (18.82)	31 (36.47)
	T4	5 (5.88)	9 (10.59)
	Tx	8 (9.41)	8 (9.41)
Node stage	N0	68 (80)	39 (45.88)	<0.001
	N1	11 (12.94)	35 (41.18)
	Nx	6 (7.06)	11 (12.94)
Distant metastasis	No	68 (80)	45 (52.94)	<0.001
	Yes	14 (16.47)	37 (43.53)
	Unknown	3 (3.53)	3 (3.53)
Age at diagnosis		37.46 ±21.48	29.09 ±18.41	0.007
Survival months		34.33 ±36.75	33.31 ±30.36	0.843

Notes: *The parametric p-value is calculated by ANOVA for numerical covariates and chi-square test for categorical covariates. P-values are printed in bold when under the 0.05 threshold.

We performed a univariate association analysis of surgery and chemotherapy with the other covariates in our dataset. This revealed that with no chemotherapy only the tumor multiplicity, size and T stage are the only covariates with significant differences between surgical treatment groups. Patients receiving no chemotherapy and no surgical treatment were noticeably older than patients receiving resection or transplantation (mean age of 53.2±20.7 Vs 29.4±17.4 and 26.7±8.1 years of age, p< 0.001) (Table 2 and ). Another interesting finding in this cohort is that pediatric and adolescent patients did not receive chemotherapy when not being resected, whereas the opposite was the case for patients with liver resection/ablation. The positivity of AFP was comparable in patients with and without metastases ().

Survival Analysis

We evaluated survival trends in the univariate analysis of both surgery and chemotherapy (Figures 1 and 2). The results have been in favor of surgery with a mean survival of 50 months. Chemotherapy alone did not show any influence on survival. Still, patients receiving both surgery and chemotherapy had a mean survival of 34 months, while patients with just resection and no chemotherapy showed a mean of 50.5 months. This finding was significant in both the chemotherapy and non –chemotherapy groups ().

Figure 1

Overall survival plot.

Figure 2

Survival plot depending on surgery, no resection and transplantation.

The 5-year overall survival (OS) rate of all patients was 40.3% while for those who had undergone surgery it was 60.7%. The overall 10-year survival was 20.8%. We examined univariate associations between each variable of interest and the time-to-event survival using Cox proportional hazards models. Out of all patients’ characteristics, young age was the significant factor for better prognosis with an HR of 0.15 (95% CI 0.06–0.35) for patients under 18 years of age ().

These results were incorporated in a multivariate Cox proportional hazards model and a Cox proportional hazards model with backward selection at a specified significance α level of 0.05.

There was no influence of gender, race, or nodal status on survival. Surgery and chemotherapy were the most significant independent prognostic factors for survival with an HR of 0.18 for resection, 0.14 for transplantation and 0.58 for chemotherapy (Table 4). Histologically undifferentiated tumors, distant metastases and age over 40 were independent negative factors of survival predictions. The factor with the highest impact on negative prognosis was the tumor stage with T4 showing an HR=5.91 (95% CI 1.19–29.33).

Table 4

Univariable Analysis and Cox Proportional Hazards Regression Analysisa

	Univariable Analysis with OS			Cox Proportional Hazards Regression Analysis with OS *,**
	HR	95% CI	P value	HR	95% CI	P value
Sex
Male	Ref			Ref
Female	1.18	0.76–1.81	0.462
Age at diagnosis			0.001			0.002
<18	Ref			Ref
18–39	1.74	0.98–3.10	0.060	1.89	1.00–3.58	0.050
40–64	2.22	1.12–4.40	0.023	3.55	1.58–8.01	0.002
>65	4.34	2.07–9.11	<0.001	5.85	2.26–15.11	<0.001
Race			0.303
Caucasian	Ref
African-American	1.35	0.73–2.51	0.337
Other	1.55	0.82–2.95	0.179
Year of diagnosis			0.422
2004–2007	Ref
2008–2011	1.21	0.76–1.93	0.418
2012–2014	0.78	0.39–1.56	0.483
Insurance status			0.953
Uninsured	Ref
Ininsured	1.11	0.15–8.06	0.915
Unknown	1.19	0.16–8.76	0.865
Tumor size			<0.001			0.083
≤5cm	Ref			Ref
>5 cm	1.02	0.57–1.84	0.965	0.97	0.45–2.10	0.931
Unknown	3.02	1.52–6.01	0.002	2.46	0.86–7.06	0.093
Tumor multiplicity			<0.001
Single lesion	Ref
Multiple lesions	2.16	1.33–3.51	0.002
Unknown	2.66	1.53–4.65	0.001
AFP status			0.003
Negative	Ref
Positive	2.26	1.39–3.67	0.001
Unknown	1.17	0.67–2.02	0.586
Tumor stage			<0.001			0.015
T1	Ref			Ref
T2	0.61	0.27–1.39	0.241	0.67	0.25–1.77	0.415
T3	2.04	1.20–3.46	0.009	2.02	1.04–4.35	0.042
T4	2.42	1.23–4.76	0.011	5.91	1.19–29.33	0.030
Tx	3.44	1.73–6.82	<0.001	1.03	0.22–4.49	0.997
Differentiation grade			0.016			0.001
G1	Ref			Ref
G2	1.32	0.52–3.39	0.559	1.73	0.62–4.79	0.292
G3	3.45	1.27–9.40	0.015	7.48	2.44–22.95	<0.001
Unknown	2.4	1.03–5.58	0.042	2.90	1.16–7.25	0.022
Node stage			0.816
N0	Ref
N1	1.09	0.68–1.75	0.714
Nx	1.25	0.59–2.63	0.557
Distant metastasis
No	Ref			Ref
Yes	2.98	1.95–4.57	<0.001	2.56	1.52–4.32	<0.001
Chemotherapy
No	Ref			Ref
Yes	1.34	0.88–2.06	0.173	0.58	0.34–0.99	0.049
Surgery or ablation			<0.001			<0.001
No surgery or ablation	Ref			Ref
Liver resection or ablation	0.17	0.11–0.26	<0.001	0.18	0.11–0.31	<0.001
Liver transplantation	0.16	0.05–0.53	0.002	0.14	0.04–0.49	0.002

Notes: aUnknown data on distant metastasis were excluded (n=6). *Number of observations in the original data set = 170. Number of observations used = 170. **Selection with an alpha level of removal of 0.05 was used. P-values are printed in bold when under the 0.05 threshold.

When utilizing a backward model of high significance with the last values as reference, the presence of distant metastases (HR=3.64; 95% CI 1.20–11.00) and the lack of surgical treatment (HR=4.64; 95% CI 1.40–15.37) were the most harmful prognostic factors. On the other hand, the age under 18 (HR 0.15; 95% CI 0.06–0.35) and under 35 (HR=0.30; 95% CI 0.14–0.64) as well as a T2 stage (HR=0.31; 95% CI 0.11–0.87) had the highest positive factor predictability. A good (G1) histological differentiation (HR=0.39; 95% CI 0.16–0.98) was also marginally an independent predictive factor of positive prognostic value while a G3 differentiation (HR=2.45; 95% CI 1.20–5.01) of negative one. Based on these findings, we created a survival nomogram with validation in 1-, 3- and 5-year survival (Figures 3 and 4). The nomogram showed that age was the largest predictor of OS, followed by surgery and tumor stage.

Figure 3

Predictive survival nomogram.

Figure 4

Predicted 1-year (A), 3-year (B) and 5-year (C) survival based on the nomogram.

Discussion

FLC is a disease with unique molecular characteristics affecting children and young adults, with a slow progression.6,23–26 Recent studies have identified a fusion gene that could provide with liquid biopsy alternatives or even therapeutic targets in the future.11,27,28 However, despite higher diagnostic rates and better understanding in its pathophysiology, FLC remains a disease with enigmatic prognosis. At present, surgical resection remains the main aim of treatment with a reported OS of 75–88 months.1 Our study showed that resection offers a median OS of 92 months with a 5-year overall survival rate of 60.7%. A very interesting finding proved the difference in survival among cases with a combination of chemotherapy and transplantation (27.44 months) versus cases treated with transplantation only (26.67 months), a finding that has not been described in extent before. Patients not undergoing surgery and not receiving treatment with chemotherapy showed a mean survival of 5.9 months, significantly lower than patients with chemotherapy. These were the only two groups with significant differences in patients’ mean age. This is an interesting finding that hints that patients who are not staged timely for chemotherapy tend to be older and survive much less. In a univariate association between chemotherapy and distant metastases, the only significant association was the age difference between patients who received chemotherapy while having a positive metastasis status (mean age 24.6 ±12.7 vs 38.1± 20.7; p=0.007); nonetheless, the survival of patients with distant metastases was comparable regardless of receiving chemotherapy treatment. This finding is significant when comparing metastasis and age in other cancer groups, where the mean age of oncological treatment is much higher.

The incidence of FLC has been previously described as low as 0.2/100,000 and previous publications on the SEER database have been reporting variable annual incidence.14,29,30 In our analysis, where patients were stratified in four categories we accounted for more than 30 new cases per year.

The already described factors predicting OS and DFS in FLC have been quite variable during the past years since most publications are based on case series with small sample size.31 Matching previous data, metastases and the differentiation of tumors were negative predictive factors. Furthermore, age and the lack of surgical treatment were factors with negative predictive value. Our most impressive finding was that larger tumors had a negative prognostic value, although one can see that most of these patients were surgically treated. This can be a case of small sample size (5 patients in total) but also a situation where recurrence occurred.

Concerning the diagnosis of FLC, multimodal imaging is of paramount importance since the clinical symptoms are inconstant and the diagnostic value of AFP is marginal. Our study confirmed that the levels of AFP are not indicative of disease or even of distant metastases. However, the null association of AFP with distant metastases at baseline in a univariate analysis has been demonstrated for the first time in this study. This finding may have more than the hypothesized limited clinical significance. A combination of this marker with other biomarkers should be tried in larger datasets.

The value of aggressive surgery has been demonstrated since the first publications on FLC and although even high-volume centers report a significant proportion of recurrent disease, the 5-year prognosis remains higher than this of not resected/ablated tumors. Our study was consistent with the findings of other analyses, which showed better survival for surgery/ablation, but in the present study, the effect of chemotherapy in combination with surgery showed actually a lower OS. This may be attributable to advanced disease, because due to age of diagnosis, one can postulate that the patients had few underlying conditions.

This study has a number of limitations apart from possible selection bias, with the database-derived factors being the most important ones. We had a relatively large sample size; however, due to the fact that other databases are not open for validation, we could not validate the analysis presented here. Another limitation is that some transplantation cases were not classified as patients who had received chemotherapy before or after transplantation, which would be a novel approach to this disease. Furthermore, the SEER does not contain data on recurrence, which would help specify our findings on the role of surgery or transplantation on survival. Finally, data on chemotherapy type/duration are required in order to assess its influence on survival.

Conclusions

The prognosis of FLC depends greatly on its surgical management. Surgery irrespective of combination with (neo-) adjuvant chemotherapy offers significantly better survival than systemic therapy alone.