Clinical Biochemistry 52 (2018) 20–25

Contents lists available at ScienceDirect

Clinical Biochemistry
journal homepage: www.elsevier.com/locate/clinbiochem

Diagnostic value of alpha-1-fetoprotein (AFP) as a biomarker for T

hepatocellular carcinoma recurrence after liver transplantation
Aventinus Nörthena,⁎, Thomas Asendorfb, Philip D. Walsonc, Michael Oellerichc
a
Department of Diagnostic and Interventional Radiology, Hannover Medical School, Carl-Neuberg-Straße 1, 30625 Hannover, Germany
b
Department of Medical Statistics, University Medical Center Göttingen, Humboldtallee 32, 37073 Göttingen, Germany
c
Department of Clinical Pharmacology, University Medical Center Göttingen, Kreuzbergring 36, 37075 Göttingen, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: Background: Alpha-1-fetoprotein (AFP) is used to monitor progression, evaluate response to therapy and predict
Alpha-1-fetoprotein recurrence of hepatocellular carcinoma (HCC) in liver transplantation (LTx) patients. To date, the diagnostic
Cut-off values value of serum AFP determinations for detecting tumor recurrence in HCC patients after LTx is unclear.
Recurrence Objective: A retrospective, single-center, cross-sectional, non-interventional study was performed with the ob-
Liver transplantation
jective of determining post-transplant cut-off AFP values for detecting HCC recurrence post LTx.
Hepatocellular carcinoma
Methods: Using receiver operating characteristic (ROC) analyses, post-transplant serum AFP values were eval-
uated against HCC recurrences in 63 HCC patients who had LTx between November 1995 and December 2011 at
the University Medical Center Göttingen (UMG). Optimal and application-independent cut points for predicting
tumor recurrence at 1, 3, and 5 years after LTx were determined.
Results: Post-LTx serum AFP was found to represent an independent risk factor (predictor) for HCC relapse. Post-
operative AFP cut-off values of 7 μg/l, 6 μg/l, and 6 μg/l, respectively, were determined to be optimal at 1, 3,
and 5 years after LTx respectively for predicting a HCC relapse. Using these cut-off values, patients were cor-
rectly classified as relapse-positive with a diagnostic sensitivity of 79%, 81%, and 77%, and as relapse-free with a
specificity of 82%, 79%, and 69%. The diagnostic accuracy measured by area under the curve (AUC) values
ranged from 0.813 to 0.886. However, a limitation is that at a clinically relevant specificity of ≥ 95%, the
analyses showed sensitivity values of only 50%, 52%, and 50%, respectively.
Conclusion: Post-transplant serum AFP may have diagnostic value to detect HCC recurrence after LTx.

1. Introduction
Hepatocellular carcinoma (HCC) tumors are highly aggressive, as-
sociated with rapid disease progression and limited therapeutic options
[1,2]. Globally, HCC is the fifth most frequent malignancy, with ap-
proximately 782,000 new cases per year, and is the second most
common cause of tumor-associated death [3,4]. In 2012, 746,000 pa-
tients died from this disease [4]. HCC is often fatal within 6 months if
left untreated as a result of extensive metastases with vascular invasion,
as well as intrahepatic, extrahepatic, and lymphatic involvement [5–7].
Since HCC occurs in the presence of cirrhosis, liver transplantation
(LTx) currently represents the optimal therapeutic approach in its early
stages, before the disease has progressed significantly [7–9]. According
to the guidelines published by the European Association for the Study of
the Liver (EASL) and the European Organisation for Research and
Treatment of Cancer (EORTC), LTx is indicated when the Milan criteria
are met (1 tumor of up to 5 cm, or a maximum of 3 tumors, each not
larger than 3 cm) [7]. The overall HCC recurrence risk after LTx in all
patients is approximately 20% [9], and, even when the Milan criteria
are followed, the rate of relapse is still between 8% and 15% [7,10].
Serum alpha-1-fetoprotein (AFP) concentrations are considered to
be helpful for monitoring HCC progression and selecting therapy after

Abbreviations: AFP, alpha-1-fetoprotein; a, year(s); avg., average; CI, confidence interval; CMIA, chemiluminescence micro-particle immunoassay; ctDNA, circulating cell-free tumor
DNA; d, day(s); diam., diameter; EASL, European Association for the Study of the Liver; EC, ethics committee; ECLIA, electro-chemiluminescence immunoassay; EORTC, European
Organisation for Research and Treatment of Cancer; et al., and others; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; LL, liver lobe; LTx, liver trans-
plantation; MVI, micro-vascular invasion; n, number; NCCN, National Comprehensive Cancer Network; npv, negative predictive value; ppv, positive predictive value; r, correlation
coefficient; ROC, receiver operating characteristic; SD, standard deviation; SEN, sensitivity; SPE, specificity; TACE, transarterial chemoembolization; UICC, Union for International Cancer
Control; UMG, University Medical Center Göttingen
⁎
Corresponding author.
E-mail addresses: aventinus.noerthen@gmx.de (A. Nörthen), thomas.asendorf@med.uni-goettingen.de (T. Asendorf), pwalson1@aol.com (P.D. Walson),
michael.oellerich@med.uni-goettingen.de (M. Oellerich).

http://dx.doi.org/10.1016/j.clinbiochem.2017.10.011
Received 13 July 2017; Received in revised form 18 September 2017; Accepted 15 October 2017
Available online 17 October 2017
0009-9120/ © 2017 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
A. Nörthen et al.
LTx [11,12]. A decrease in serum AFP values to < 20 μg/l within
2 months was reported to be indicative of complete tumor removal [9],
whereas AFP levels that remain elevated after LTx or that initially fall to
normal but then again increase have been reported to be indications of
a residual tumor or tumor recurrence, respectively [13,14]. However,
the diagnostic value of post-LTx serum AFP values for early detection of
tumor recurrence has not yet been either conclusively demonstrated or
sufficiently evaluated to date. Neither the European EASL-EORTC
guidelines nor the American guidelines published by the National
Comprehensive Cancer Network (NCCN) currently recommend the use
of any specific AFP cut point to detect HCC recurrence post-LTx [7,12].
This study investigated the diagnostic value of serum AFP con-
centrations as a biomarker of post-LTx tumor recurrence in HCC pa-
tients.
2. Materials and methods
2.1. Patient selection
The study population consisted of 63 Caucasian HCC patients who
had orthotopic liver transplantations with allogeneic liver organs from
deceased donors at the University Medical Center Göttingen (UMG)
between November 1995 and December 2011. HCC patients were
identified from a review of the UMG LTx medical records of 850 pa-
tients who had received a LTx during the period between 1st January
1993 and 31st December 2011. It was possible to track the progression
of HCC in 101 of these LTx patients using the data contained in these
medical records. However, only those HCC patients (n = 63) without
mixed carcinomas (hepato- or cholangiocellular) and who had at least
one post-LTx serum AFP concentration available were included in this
population analyses.
The research study was reviewed and approved by the ethics com-
mittee (EC) of the UMG (file number: DOK_73_2012). All data included
were collected solely for clinical purposes and were previously anon-
ymized. Prior, written informed consent for all routine diagnostic and
medical treatment procedures/interventions was obtained for each
patient.
2.2. Study data collection
The relevant tumor-related medical history of each patient included
in the study population was tracked from their initial presentation to
UMG until February 2013. In addition to the post-LTx serum AFP
concentration measurements, any data recorded between initial contact
and tumor recurrence, death, or the final follow-up appointment at
UMG prior to or during February 2013 were included in the review. The
medical history could only be reliably tracked for follow-up care pro-
vided at UMG. Data from externally performed follow-up examinations
were only included in the analyses if recorded in the patient's UMG
medical record. For those patients who transferred their post-LTx
follow-up care elsewhere than UMG before the end of the study
(February 2013), only data recorded in their UMG medical records
prior to transfer were included in the analyses. In such cases, the date of
the last recorded contact with the patient in the UMG medical record
was defined as the end of their follow-up care.
The parameters evaluated for this study, including tumor histo-
pathology, serum AFP values, disease recurrence status, and survival
data, were taken from both the UMG LTx database and the internal
hospital information system. Some additional non-digital information
was also available from the patients' medical records.
2.3. Methods used to determine serum AFP
The AFP measurements were all routine assays performed in the
UMG Clinical Chemistry laboratory; initially (until 2012) using the
cobas® 8000 modular analyzer series system along with the cobas e 602
21
Clinical Biochemistry 52 (2018) 20–25
module from Roche. This system uses electro-chemiluminescence im-
munoassays (ECLIA) according to the manufacturers' instructions to
determine AFP levels. In 2012, this equipment was replaced by the
ARCHITECT i SYSTEM from Abbott, which determines AFP con-
centrations using chemiluminescence micro-particle immunoassays
(CMIA) according to the manufacturer's specifications. The lower limit
of detection was 0.61 ng/ml (0.61 μg/l) for the cobas e 602 test kits,
and 1 ng/ml (1 μg/l) for the ARCHITECT AFP assays. Direct compar-
isons of the AFP results obtained in 39 patients yielded a good agree-
ment between both methods (Passing/Bablok statistics: y = 0.972 × –
0.548; r = 0.994).
2.4. Statistical methods
For interval-scale parameters, the mean value, standard deviation,
and range of each parameter were determined. For parameters with
skewed distributions, the median, 25th percentile, and 75th percentile
were calculated. For nominal-scale parameters, the number of times
that each value occurred was recorded, as well as the percentage of the
total number of values.
A univariate Cox regression was performed to examine the asso-
ciation between the AFP values and the recurrence risk, followed by a
multivariate Cox regression in order to determine the additional in-
formation provided by the AFP in the presence of other parameters. For
both Cox regressions, the hazard ratio (HR) was determined along with
a 95% confidence interval (95% CI).
The metrics used to describe the prognostic and diagnostic quality
of each potential cut point were the area under the curve (AUC), the
sensitivity (SEN), and the specificity (SPE) which were calculated using
receiver operating characteristic (ROC) analyses, as was the positive
(ppv) and negative predictive values (npv) with 95% confidence in-
tervals [15]. The sensitivity and specificity were simultaneously max-
imized in order to determine the optimal AFP cut points [16]. To
evaluate the diagnostic value of post-LTx serum AFP values, the sensi-
tivity at an application-specific specificity of ≥95% was also de-
termined.
The distribution of the AFP values showed significant right skew.
Before performing the Cox regressions, the AFP values were converted
to logs to meet the conditions for applying the procedure. However, in
order to simplify interpretations, the non-log values were used for the
AFP when expressing the optimal cut points and calculating the quality
metrics. ROC analyses were performed with the patients' maximum
serum AFP levels measured within 1, 3 and 5 years after LTx, respec-
tively. Maximum AFP levels that reached or exceeded a respective cut
point were classified as relapse-positive and maximum AFP levels lower
than a respective cut point were classified as relapse-free. To prevent
potential selection bias as much as possible, no minimum time interval
was required between LTx and AFP measurements in order for the AFP
levels to be included in the statistical analyses.
Statistical analysis and graphical rendering were performed using
the statistical software R® for Windows, Version 3.2.0 [17].
3. Results
3.1. Patient demographics
There were 53 (84.1%) male and 10 (15.9%) females in the study
popoulation. All transplant procedures were performed orthotopically
(after hepatectomy) using organs from deceased donors. In 58 cases
(92.1%), complete organs were used, an additional 5 (7.9%) were
partial liver transplants. A total of 746 post-LTx serum AFP measure-
ments were available for inclusion in the analyses. The demographic
composition of the population, including tumor histopathology, AFP
values, and disease recurrence status, are presented in Tables 1 and 2.
A. Nörthen et al. Clinical Biochemistry 52 (2018) 20–25

Table 1
Population data (demographics, etiology, TACE, and tumor histopathology).

Variable Data

Population size [n (%)] 63 (100)

Gender [n (%)]
Male/female 53 (84.1)/10 (15.9)
Race [n (%)]
Caucasian 63 (100)
Avg. age of recipient ± SD (range) [a] 59.6 ± 10.5 (30.5–76.5)
Type of LTx [n (%)]
Complete LTx/split LTx 58 (92.1)/5 (7.9)
Cirrhosis [n (%)]
No/yes 3 (4.8)/60 (95.2)
Cirrhosis etiology [n (%)]a
HBV/HCV/HBV + HCV 5 (7.9)/20 (31.7)/3 (4.8)
Alcohol/hemochromatosis/cryptogenic 21 (33.3)/3 (4.8)/8 (12.7)
Bridging with TACE
No/yes [n (%)] 35 (55.6)/28 (44.4)
Avg. therapy cycles ± SD (range) 3.6 ± 2.4 (1–9)
Number of tumors [n (%)]a
Solitary/multiple 23 (36.5)/38 (60.39)
Avg. max. diam. of largest tumor ± SD (range) [cm] 4.5 ± 3.9 (0.4–17)
Tumor location within lobes [n (%)]a
Left LL/right LL/both lobes 10 (15.9)/26 (41.3)/25 (39.7)
MVI [n (%)]a
No/yes 45 (74.6)/16 (25.4)
Grading [n (%)]a
G1/G1–2 7 (11.1)/2 (3.2)
G2/G3 48 (76.2)/4 (6.3)
UICC stage [n (%)]a
I/II 19 (30.2)/17 (27)
IIIa/IIIb/IIIC 6 (9.5)/11 (17.5)/4 (6.3)
IVa/IVb 3 (4.8)/1 (1.6)
Milan criteria [n (%)]a
Not met/met 37 (58.7)/24 (38.1)

n: number of patients, avg.: average, a: year(s), max. diam.: maximum, Diameter, LL: liver lobe, MVI: micro-vascular
invasion.
a
n = 61 (because parameter was unavailable or unknown in 2 patients).

Table 2 3.2. Post-LTx serum AFP values: univariate and multivariate analyses
Population data (AFP, recurrence, follow-up, and Exitus letalis).
According to the results of univariate Cox regression, post-LTx AFP
Variable Data
concentrations had a highly significant impact on the risk of relapse
Total post-LTx AFP measurements 746 [n = 63; HR = 1.335 (95% CI: 1.194–1.492), p < 0.0001].
Post-transplant test frequency per patient The multivariate Cox regression analysis confirmed that the post-
25th percentile/median/75th percentile 4/8/15 LTx serum AFP levels represent an independent risk factor (predictor)
Post-LTx AFP levels
25th percentile/median/75th percentile [μg/l] 3/5/15.5
of post-transplant recurrence [n = 58; HR = 1235 (95% CI:
Minimum/maximum [μg/l] 2/191,600 1.079–1.415); p < 0.05]. Besides the post-LTx AFP values, the number
Time interval between last AFP measurement and of tumors, tumor size, location within the liver lobes, and tumor mi-
LTx crovascular invasion were additional predictors, as were the Milan
25th percentile/median/75th percentile [a] 0.7/1.25/3.84
criteria scores.
Recurrence [n (%)] no/yes 40 (31.7)/23 (36.5)
Type of recurrence condition [n (%)]
Intrahepatic/extrahepatic/intra- and 5 (7.9)/9 (14.3)/9 (14.3)
extrahepatic 3.3. Post-LTx AFP cut points for predicting recurrence
Location of extrahepatic recurrences [n (%)]
Lung/lymph nodes 10 (15.9)/4 (6.3)
Bones/peritoneum/endovascular around portal 3 (4.8)/2 (3.2)/2 (3.2)
The optimal and application-independent serum AFP cut-off values
vein for predicting recurrence at 1, 3, and 5 years after LTx were determined
Post-transplant follow-up to be 7 μg/l, 6 μg/l, and 6 μg/l, respectively (Fig. 1a–c). These cut
25th percentile/median/75th percentile [a] 0.7/1.8/4.0 points correctly classified HCC patients as relapse-positive with a sen-
Died [n (%)] no/yes 40 (31.7)/23 (36.5)
sitivity of 79%, 81%, and 77% and as relapse-free with a specificity of
Cause of death [n (%)]
Relapse/sepsis 14 (23.8)/4 (6.3) 82%, 79%, and 69% respectively. The diagnostic accuracy measured by
pneumonia/CAD/laryngeal ca./esophageal ca. 1 (1.6)/1 (1.6)/1 (1.6)/1 AUC values ranged from 0.813 to 0.886 (Table 3).
(1.6) At a clinically relevant specificity of ≥95%, post-LTx serum AFP
levels at 1, 3, and 5 years after LTx showed sensitivity values of 50%,
SD: standard deviation, a: year(s), d: day(s), n: number of patients, pat.: patients, ca.:
52%, and 50%, respectively for prediction of recurrence. Post-trans-
carcinoma.
plant AFP cut points with specificity ≥ 95% to predict recurrence (ROC
analyses 1, 3 and 5 years after LTx) were much higher; 354 μg/l,
223 μg/l and 223 μg/l respectively (Table 4).

22
A. Nörthen et al. Clinical Biochemistry 52 (2018) 20–25

Fig. 1. a–c Post-transplant AFP cut points 1 (1a), 3 (1b) and 5 (1c) years after LTx (ROC curves).
The optimal AFP cut points are marked with an asterisk (*). For illustrative purposes, the cut points are (only) shown as the transformed to log form in the ROC graphs. The logarithm can
be reverted by applying the exponential function (ex). Example: For an AFP cut point of 3 μg/l, log (3 μg/l) ≈ 1.098 log-μg/l and e1.098 log-μg/l ≈ 3 μg/l.

3.4. Increase in post-LTx AFP values for predicting recurrence
A distinct increase in post-LTx serum AFP levels during any 90 days
post-LTx period was associated with a high probability of recurrence
within 5 years. According to the ROC analysis for predicting recurrence
5 years after LTx, the optimal value for the post-LTx 90-day increase in
serum AFP values was 6 μg/l (Fig. 2). This value had a sensitivity of
68% and a specificity of 77% (Table 5).
At a clinically relevant specificity of ≥ 95%, the analysis showed an
application-specific sensitivity of only 50%. The post-LTx increase in
AFP values over any 90 day period with specificity ≥ 95% (ROC ana-
lysis for predicting recurrence 5 years after LTx) was much higher;
165 μg/l (Table 6).
4. Discussion
The diagnostic value of post-LTx serum AFP levels for detecting the
recurrence of HCC after LTx was investigated. Post-LTx AFP
concentrations were found to represent an independent risk factor for
prediction of HCC recurrence. Post-LTx serum AFP values of 7 μg/l,
6 μg/l, and 6 μg/l were determined to be the optimal cut-off values 1, 3,
and 5 years after LTx, respectively for predicting recurrence. These cut-
off values correctly classified patients as relapse-positive with a sensi-
tivity of 79%, 81%, and 77%, and as relapse-free with a specificity of
82%, 79%, and 69%. While these cut points lie within the reported
reference range (≤ 10 μg/l) used to evaluate concentrations in pregnant
women it should be noted that median (SD) AFP concentrations as low
as 3.04 μg/l ( ± 1.9) have been reported in non-pregnant adults [18].
Pre-LTx serum AFP has been extensively studied either alone or in
combinations as a biomarker of overall survival or to predict post LTx
HCC recurrence [9,19,20]. However, while used by some to monitor
recurrence, there have been few previous studies investigating the di-
agnostic value of post-LTx serum AFP values for detecting the recur-
rence of HCC [21]. The quality metrics and AFP cut points determined
by two previous studies are broadly consistent with the results of this
study. Chang et al. suggested a serum AFP value of ≥ 5.45 μg/l as the

Table 3
Optimal post-transplant AFP cut points for predicting recurrence (ROC analyses 1, 3 and 5 years after LTx).

Time of recurrence prediction Optimal AFP cut point AUC (95% CI) SEN (95% CI) [%] SPE (95% CI) [%] ppv (95% CI) [%] npv (95% CI) [%]

1 year after LTx (n = 53) 7 μg/l 0.886 (0.771–1.002) 79 (49–95) 82 (66–92) 61 (36–83) 91 (77–98)
3 years after LTx (n = 40) 6 μg/l 0.862 (0.744–0.98) 81 (58–95) 79 (54–94) 81 (58–95) 79 (54–94)
5 years after LTx (n = 35) 6 μg/l 0.813 (0.672–0.954) 77 (55–92) 69 (39–91) 81 (58–95) 64 (35–87)

AUC: area under ROC curve, 95% CI: 95% confidence interval, SEN: sensitivity, SPE: specificity, ppv: positive predictive value, npv: negative predictive value, n: number of patients.

23
A. Nörthen et al. Clinical Biochemistry 52 (2018) 20–25

Table 4
Post-transplant AFP cut points with specificity ≥ 95% (ROC analyses 1, 3 and 5 years after LTx).

Time of recurrence prediction AFP cut point AUC (95% CI) SEN (95% CI) [%] SPE (95% CI) [%] ppv (95% CI) [%] npv (95% CI) [%]

1 year after LTx (n = 53) 354 μg/l 0.886 (0.771–1.002) 50 (23–77) 97 (87–100) 88 (47–100) 84 (71–94)
3 years after LTx (n = 40) 223 μg/l 0.862 (0.744–0.98) 52 (30–74) 100 (75–100) 100 (62–100) 66 (46–82)
5 years after LTx (n = 35) 223 μg/l 0.813 (0.672–0.954) 50 (28–72) 100 (66–100) 100 (62–100) 54 (33–74)

AUC: area under ROC curve, 95% CI: 95% confidence interval, SEN: sensitivity, SPE: specificity, ppv: positive predictive value, npv: negative predictive value, n: number of patients.

Fig. 2. Post-transplant increase in serum AFP values 5 years after LTx (ROC curve).
The optimal AFP increase is marked with an asterisk (*).
post-operative threshold for detecting recurrence based on a population
of 64 liver resection and 8 LTx HCC patients. This cut-off value
achieved a sensitivity of 84.4%, and a specificity of 77.1% [22]. Ac-
cording to a second study in 100 HCC patients by Yamashiki et al., a
post-LTx AFP cut point of 10 ng/ml (10 μg/l) had a sensitivity of 78%
for detecting recurrence, as well as a specificity of 98% [23]. The
prognostic value of changes in AFP have also been more recently in-
vestigated [24]. Yoo et al. reported that rapid normalization of AFP
values at 1 month post-LTx was “a useful clinical marker for HCC re-
currence” [13]. However, neither the European EASL-EORTC guide-
lines nor the American NCCN guidelines recommend an AFP cut point
for detecting HCC recurrence in LTx patients [7,12].
In clinical practice, in order for a tumor marker to be considered to
be adequate for screening and early detection, it must achieve sensi-
tivity and specificity values of at least 90% [25]. At a standardized
specificity of 95%, a sensitivity of at least 50% is required [26]. In light
of these requirements for target biomarkers, the diagnostic value of
serum AFP levels measured after LTx was not ideal. The ROC analyses
1, 3, and 5 years after LTx revealed sensitivity values of only 50%, 52%,
and 50%, respectively, at a specificity of ≥ 95%. Therefore, one of the
conclusions of this study is that more efficient biomarkers must be
developed for the post-LTx care of HCC patients. One approach in-
creasingly used for preoperative risk prediction is the use of combina-
tions of biomarkers, including AFP. Examples include the MORAL score
that combines AFP values with prothrombin induced by Vitamin K
absence and a newer version called the AFP score that combines AFP
with tumor size [27,28]. Future combinations of biomarkers are likely
to include molecular markers since allelic imbalance in micro-satellites
in pre-LTx tumor DNA were shown in a recent review to be highly as-
sociated with HCC recurrence [19]. Recently, a less invasive and pos-
sibly superior to biopsy method for evaluation of tumor DNA using
circulating cell-free tumor DNA (ctDNA) has been shown to be a pro-
mising candidate molecular biomarker for this use [29–31].
The overall recurrence rate for LTx HCC patients can be up to 20%
[9]. Once the diagnosis of recurrence has been established, the survival
period is currently less than one year [11,23]. Post-LTx AFP levels with
values equal to or greater than the cut points identified in this study
were found to represent an important independent predictor of tumor
relapse. In some cases, suggestive AFP levels were observed as early as
6 months before the clinical diagnosis of recurrence was later estab-
lished (positive lead time). Practical, inexpensive biomarkers that
consistently provide positive lead times might be helpful to decide
when treatment options such as immunotherapy should be im-
plemented. This could prolong survival or even prevent tumor recur-
rence [32]. Until more effective biomarkers are established in pro-
spective, controlled, multi-center trials, application of more intensive
imaging or use of these post-LTx threshold cut-offs and distinctive in-
creases in AFP values identified in this study can be considered for use
along with other findings. Despite their non-optimal diagnostic sensi-
tivity, the time involved and financial cost of AFP monitoring are lower
than sectional imaging and, unlike CT and MRI procedures, capacity is
rarely an issue for laboratory tests [9,11].
This study has many limitations including its single-center, retro-
spective, cross-sectional, non-interventional design. The population of
only 63 Caucasian patients also limits the representativeness of the
results. The cut-off values and quality metrics need to be verified in
other, larger, multi-center, prospective, more diverse patient popula-
tions that routinely collect both clinical data and AFP values at pre-
specified time intervals. Although AFP methods are usually standar-
dized against the AFP WHO reference material, methods other than the

Table 5
Optimal post-transplant increase in AFP values over any 90 day period for predicting recurrence (ROC analysis 5 years after LTx).

Time of recurrence prediction Optimal increase AUC (95% CI) SEN (95% CI) [%] SPE (95% CI) [%] ppv (95% CI) [%] npv (95% CI) [%]
in AFP value

5 years after LTx (n = 35) 6 μg/l 0.764 (0.608–0.920) 68 (45–86) 77 (46–95) 83 (59–96) 59 (33–82)

AUC: area under ROC curve, 95% CI: 95% confidence interval, SEN: sensitivity, SPE: specificity, ppv: positive predictive value, npv: negative predictive value, n: number of patients.

Table 6
Post-transplant increase in AFP values over any 90 day period with specificity ≥95% (ROC analysis for predicting recurrence 5 years after LTx).

Time of recurrence prediction AFP value increase AUC (95% CI) SEN (95% CI) [%] SPE (95% CI) [%] ppv (95% CI) [%] npv (95% CI) [%]

5 years after LTx (n = 35) 165 μg/l 0.764 (0.608–0.920) 50 (28–72) 100 (66–100) 100 (62–100) 54 (33–74)

AUC: area under ROC curve, 95% CI: 95% confidence interval, SEN: sensitivity, SPE: specificity, ppv: positive predictive value, npv: negative predictive value, n: number of patients.

24
A. Nörthen et al.
Roche and Abbott assays might require the use of different cut-off va-
lues.
The results suggest, that post-transplant serum AFP values even
within the “normal range” may have diagnostic value as a biomarker to
detect recurrence in HCC patients after LTx. However, in patients with
non-suggestive AFP levels relapse must by no means be excluded due to
the sensitivity of only 50–52% at a clinically relevant specificity of
≥ 95%. Research into ctDNA in plasma or urine DNA could represent
new opportunities toward developing more effective target biomarkers
for detecting HCC recurrence [31,33,34]. A combination of cfDNA and
the conventional HCC biomarker AFP has been proposed to improve the
diagnostic value [35,36].
References
[1] M. Kudo, F. Trevisani, G.K. Abou-Alfa, L. Rimassa, Hepatocellular carcinoma:
therapeutic guidelines and medical treatment, Liver Cancer 6 (2016) 16–26, http://
dx.doi.org/10.1159/000449343.
[2] Z. Huang, N. Zhang, W. Li, J. Cao, L. Zhang, Y. Chen, Expression of CHODL in
hepatocellular carcinoma affects invasion and migration of liver cancer cells, Oncol.
Lett. 13 (2017) 715–721, http://dx.doi.org/10.3892/ol.2016.5466.
[3] J. Ferlay, I. Soerjomataram, R. Dikshit, S. Eser, C. Mathers, M. Rebelo, D.M. Parkin,
D. Forman, F. Bray, Cancer incidence and mortality worldwide: sources, methods
and major patterns in GLOBOCAN 2012, Int. J. Cancer 136 (2015) E359–E386,
http://dx.doi.org/10.1002/ijc.29210.
[4] World Health Organization, Liver Cancer Estimated Incidence, Mortality and
Prevalence Worldwide in 2012, http://globocan.iarc.fr/Pages/fact_sheets_cancer.
aspx, (2012) (accessed October 11, 2016).
[5] E.G. Giannini, F. Farinati, F. Ciccarese, A. Pecorelli, G.L. Rapaccini, M. Di Marco,
L. Benvegnù, E. Caturelli, M. Zoli, F. Borzio, M. Chiaramonte, F. Trevisani,
Prognosis of untreated hepatocellular carcinoma, Hepatology 61 (2015) 184–190,
http://dx.doi.org/10.1002/hep.27443.
[6] E.P. Weledji, G.E. Orock, M.N. Ngowe, D.S. Nsagha, How grim is hepatocellular
carcinoma? Ann. Med. Surg. 3 (2014) 71–76, http://dx.doi.org/10.1016/j.amsu.
2014.06.006.
[7] European Association for the Study of the Liver, European Organisation for
Research and Treatment of Cancer, EASL-EORTC Clinical Practice Guidelines:
management of hepatocellular carcinoma, J. Hepatol. 56 (2012) 908–943, http://
dx.doi.org/10.1016/j.jhep.2011.12.001.
[8] J. de V. de Goyet, R.L. Meyers, G.M. Tiao, B. Morland, Beyond the Milan criteria for
liver transplantation in children with hepatic tumours, Lancet Gastroenterol.
Hepatol. 2 (2017) 456–462, http://dx.doi.org/10.1016/S2468-1253(17)30084-5.
[9] B. Charrière, C. Maulat, B. Suc, F. Muscari, Contribution of alpha-fetoprotein in
liver transplantation for hepatocellular carcinoma, World J. Hepatol. 8 (2016)
881–890, http://dx.doi.org/10.4254/wjh.v8.i21.881.
[10] V. Mazzaferro, E. Regalia, R. Doci, S. Andreola, A. Pulvirenti, F. Bozzetti,
F. Montalto, M. Ammatuna, A. Morabito, L. Gennari, Liver transplantation for the
treatment of small hepatocellular carcinomas in patients with cirrhosis, N. Engl. J.
Med. 334 (1996) 693–699, http://dx.doi.org/10.1056/NEJM199603143341104.
[11] P.-A. Clavien, M. Lesurtel, P.M.M. Bossuyt, G.J. Gores, B. Langer, A. Perrier,
Recommendations for liver transplantation for hepatocellular carcinoma: an inter-
national consensus conference report, Lancet Oncol. 13 (2012) e11–e22, http://dx.
doi.org/10.1016/S1470-2045(11)70175-9.
[12] National Comprehensive Cancer Network, NCCN clinical practice guidelines in
oncology (NCCN guidelines®): hepatobiliary, Cancers (2015).
[13] T. Yoo, K.-W. Lee, N.-J. Yi, Y.R. Choi, H. Kim, S.-W. Suh, J.H. Jeong, J.-M. Lee, K.-
S. Suh, Peri-transplant change in AFP level: a useful predictor of hepatocellular
carcinoma recurrence following liver transplantation, J. Korean Med. Sci. 31 (2016)
1049–1054, http://dx.doi.org/10.3346/jkms.2016.31.7.1049.
[14] L. Liu, Y. Zhao, J. Jia, H. Chen, W. Bai, M. Yang, Z. Yin, C. He, L. Zhang, W. Guo,
J. Niu, J. Yuan, H. Cai, J. Xia, D. Fan, G. Han, The prognostic value of alpha-
fetoprotein response for advanced-stage hepatocellular carcinoma treated with
sorafenib combined with transarterial chemoembolization, Sci. Rep. 6 (2016)
19851, http://dx.doi.org/10.1038/srep19851.
[15] D.L. Riddle, P.W. Stratford, Interpreting validity indexes for diagnostic tests: an
illustration using the berg balance test, Phys. Ther. 79 (1999) 939–948.
[16] R.J. Gallop, P. Crits-Christoph, L.R. Muenz, X.M. Tu, Determination and inter-
pretation of the optimal operating point for ROC curves derived through general-
ized linear models, Underst. Stat. 2 (2003) 219–242, http://dx.doi.org/10.1207/
25
Clinical Biochemistry 52 (2018) 20–25
S15328031US0204_01.
[17] R Core Team, R, A Language and Environment for Statistical Computing, Vienna,
Austria, https://www.R-project.org/, (2016).
[18] D. Ball, E. Rose, E. Alpert, Alpha-fetoprotein levels in normal adults, Am. J. Med.
Sci. 303 (1992) 157–159.
[19] H.-C. Pommergaard, J. Burcharth, J. Rosenberg, A. Rasmussen, Serologic and mo-
lecular biomarkers for recurrence of hepatocellular carcinoma after liver trans-
plantation: a systematic review and meta-analysis, Transplant. Rev. 30 (2016)
171–177, http://dx.doi.org/10.1016/j.trre.2016.03.001.
[20] A. Notarpaolo, R. Layese, P. Magistri, M. Gambato, M. Colledan, G. Magini,
L. Miglioresi, A. Vitale, G. Vennarecci, C.D. Ambrosio, P. Burra, F. Di Benedetto,
S. Fagiuoli, M. Colasanti, G. Maria Ettorre, A. Andreoli, U. Cillo, A. Laurent,
S. Katsahian, E. Audureau, F. Roudot-Thoraval, C. Duvoux, Validation of the AFP
model as a predictor of HCC recurrence in patients with viral hepatitis-related
cirrhosis who had received a liver transplant for HCC, J. Hepatol. 66 (2017)
552–559, http://dx.doi.org/10.1016/j.jhep.2016.10.038.
[21] L. dos S. Schraiber, A.A. de Mattos, M.L. Zanotelli, G.P.C. Cantisani, A.B. de
M. Brandão, C.A. Marroni, G. Kiss, L. Ernani, P. dos S. Marcon, Alpha-fetoprotein
level predicts recurrence after transplantation in hepatocellular carcinoma, Med.
(Baltimore) 95 (2016) e2478, http://dx.doi.org/10.1097/MD.0000000000002478.
[22] S.K. Chang, W.W. Hlaing, R.Q. Yu, T.W. Lee, I.S. Ganpathi, K.K. Madhavan, Value of
alpha-foetoprotein for screening of recurrence in hepatocellular carcinoma post
resection, Singap. Med. J. 53 (2012) 32–35.
[23] N. Yamashiki, Y. Sugawara, S. Tamura, R. Tateishi, H. Yoshida, J. Kaneko,
Y. Matsui, J. Togashi, M. Akahane, M. Makuuchi, M. Omata, N. Kokudo,
Postoperative surveillance with monthly serum tumor markers after living-donor
liver transplantation for hepatocellular carcinoma, Hepatol. Res. 40 (2010)
278–286, http://dx.doi.org/10.1111/j.1872-034X.2009.00591.x.
[24] D. Nobuoka, Y. Kato, N. Gotohda, S. Takahashi, T. Nakagohri, M. Konishi,
T. Kinoshita, T. Nakatsura, Postoperative serum α-fetoprotein level is a useful
predictor of recurrence after hepatectomy for hepatocellular carcinoma, Oncol.
Rep. 24 (2010) 521–528.
[25] V. Brower, Biomarkers: portents of malignancy, Nature 471 (2011) S19–S21,
http://dx.doi.org/10.1038/471S19a.
[26] R. Lamerz, General criteria for tumor marker determination in laboratory and
clinical application, Lab. Med. 16 (1992) 13–20.
[27] P.A. Clavien, P. Dutkowski, K.D. Lillemoe, Defining MoRAL after liver transplan-
tation, Ann. Surg. 265 (2017) 555–556, http://dx.doi.org/10.1097/SLA.
0000000000002008.
[28] K.J. Halazun, M. Najjar, R.M. Abdelmessih, B. Samstein, A.D. Griesemer,
J.V. Guarrera, T. Kato, E.C. Verna, J.C. Emond, R.S. Brown, Recurrence after liver
transplantation for hepatocellular carcinoma: a new MORAL to the story, Ann. Surg.
265 (2017) 557–564, http://dx.doi.org/10.1097/SLA.0000000000001966.
[29] S.-J. Dawson, D.W.Y. Tsui, M. Murtaza, H. Biggs, O.M. Rueda, S.-F. Chin,
M.J. Dunning, D. Gale, T. Forshew, B. Mahler-Araujo, S. Rajan, S. Humphray,
J. Becq, D. Halsall, M. Wallis, D. Bentley, C. Caldas, N. Rosenfeld, Analysis of cir-
culating tumor DNA to monitor metastatic breast cancer, N. Engl. J. Med. 368
(2013) 1199–1209, http://dx.doi.org/10.1056/NEJMoa1213261.
[30] E. Schütz, M.R. Akbari, J. Beck, H. Urnovitz, W.W. Zhang, K. Bornemann-Kolatzki,
W.M. Mitchell, R.K. Nam, S.A. Narod, Chromosomal instability in cell-free DNA is a
serum biomarker for prostate cancer, Clin. Chem. 61 (2015) 239–248, http://dx.
doi.org/10.1373/clinchem.2014.226571.
[31] M. Oellerich, E. Schütz, J. Beck, P. Kanzow, P.N. Plowman, G.J. Weiss, P.D. Walson,
Using circulating cell-free DNA to monitor personalized cancer therapy, Crit. Rev.
Clin. Lab. Sci. 54 (2017) 205–218, http://dx.doi.org/10.1080/10408363.2017.
1299683.
[32] W. Ma, L. Wu, F. Zhou, Z. Hong, Z. Liu, Y. Yuan, T cell–associated immunotherapy
for hepatocellular carcinoma, Cell. Physiol. Biochem. 41 (2017) 609–622, http://
dx.doi.org/10.1159/000457883.
[33] H.-W. Hann, S. Jain, G. Park, J.D. Steffen, W. Song, Y.-H. Su, Detection of urine
DNA markers for monitoring recurrent hepatocellular carcinoma, Hepatoma Res. 3
(2017) 105, http://dx.doi.org/10.20517/2394-5079.2017.15.
[34] J. Lou, L. Zhang, S. Lv, C. Zhang, S. Jiang, Biomarkers for hepatocellular carcinoma,
biomark, Cancer 9 (2017) 1–9, http://dx.doi.org/10.1177/1179299X16684640.
[35] W. Liao, Y. Mao, P. Ge, H. Yang, H. Xu, X. Lu, X. Sang, S. Zhong, Value of quan-
titative and qualitative analyses of circulating cell-free DNA as diagnostic tools for
hepatocellular carcinoma, Medicine (Baltimore) 94 (2015), http://dx.doi.org/10.
1097/MD.0000000000000722.
[36] M. Shi, M.-S. Chen, K. Sekar, C.-K. Tan, L.L. Ooi, K.M. Hui, A blood-based three-
gene signature for the non-invasive detection of early human hepatocellular car-
cinoma, Eur. J. Cancer 50 (2014) 928–936, http://dx.doi.org/10.1016/j.ejca.2013.
11.026.