Key words circulating tumor cells - CTC - thyroid cancer - tumor marker
Introduction
Radioiodine therapy (RIT) with I-131 is a standard method for the treatment of differentiated
thyroid carcinoma, enabling the successful ablation of thyroid carcinoma cells. As
a systemic form of treatment, RIT has the major advantage of reducing the risk of
local recurrences as well as distant metastases [1 ]. RIT can also serve as diagnostic tool, allowing the visualization of normal and
carcinomatous thyroid tissue by means of whole-body scintigraphy. Furthermore, the
complete ablation of normal thyroid tissue via targeting of iodine-avid cells allows
to use the increase of the thyroid-specific protein thyroglobulin as a tumor marker
[2 ].
The success of RIT, whether as first or repeated therapy, is usually monitored within
3–6 months after therapy by means of whole-body scintigraphy, ultrasound scan of the
neck region, and serum levels of serum thyroglobulin (sTg) [3 ]. Due to the protracted efficacy and long half-life of I-131 (8.02 days), clinical
manifestations of therapy effects cannot be expected at earlier time points. Therefore,
there is a need to identify candidate markers for an early prediction of the therapeutic
response to RIT. In some tumors, the identification of circulating tumoral cells in
the blood may represent a potential marker of therapy response.
Metastases are the result of systemic dissemination of cancer. The process of metastatic
spread consists of entry of carcinoma cells in blood circulation, followed by egress
of cells from circulation into the sites of metastatic formation [4 ]. The epithelial cell adhesion molecule EpCAM (CD326) has been found to be overexpressed
not only locally in the tissue of nearly all types of carcinomas, but also in circulating
tumoral cells originating from the primary tumor [5 ]
[6 ]
[7 ]. In breast and bronchial carcinoma, for example, EpCAM-positive circulating epithelial
cells (CEC) can be identified by means of fluorescence-labeled antibodies directed
against the membrane-bound EpCAM; also, the presence and increase of these cells in
the blood of patients with carcinoma can be interpreted as a sign of recurrence [8 ]
[9 ]
[10 ]. However, the demonstration of CEC is not carcinoma-specific, since these cells
can also be found in benign diseases [11 ].
In a previous study, we have demonstrated that CEC can be found in patients with different
thyroid diseases, and that the number of CEC was particularly elevated in patients
with differentiated thyroid carcinoma who had undergone recent thyroidectomy [12 ]. However, only a single point in time was studied in each patient group, therefore
there was no information on the time course of the therapy-induced changes. Indeed,
studies in other carcinoma entities have shown that the time course of CEC levels
during antineoplastic treatment correlates with the clinical response to treatment
[8 ]
[13 ].
The present study was undertaken to investigate whether and how the numbers of CEC
respond to RIT in patients with differentiated thyroid carcinoma. 2 types of patients
were considered for this study, i. e., those undergoing a first RIT after thyroidectomy
(thus having thyroid remnants primarily of non-malignant nature) and those undergoing
a repeated RIT because of tumor recurrences (thus having thyroid tissue of predominantly
carcinomatous nature). 2 early time points post RIT (day 2 and day 14) and one later
time point (3 months), the latter in parallel with routine clinical assessment of
therapy response, were chosen for comparison to pretreatment levels.
Materials and Methods
28 patients were enrolled in the study after approval of the local Ethics Committee.
The patients were divided in 2 groups: 1) Patients recently subjected to thyroidectomy
(4–6 weeks) and with histological diagnosis of differentiated thyroid carcinoma (DTC);
these patients underwent a first RIT with activities between 2 and 11 GBq (54–297 mCi),
in order to ablate the residual thyroid tissue (mostly non-carcinomatous) (RITfirst , n=13); and 2) Patients with evidence/persistence of differentiated thyroid carcinoma
tissue, due to which they received a repeated RIT with activities between 5.5 and
15 GBq (RITrep , n=15). In the RITfirst group, patients of different TNM stage were included, ranging from T1aN0M0 to T4N1bM1.
In the RITrep group, 2 patients were initially staged T1N0M0, the remaining patients’ staging were
T2 or 3 N0 (except for 1 patient) M0 (except for 1 patient).
The number of CEC was assessed at 4 time points: the day before the administration
of I-131 (baseline value), 2 and 14 days post treatment, and finally 3 months afterwards
([Fig. 1 ]). Blood samples were collected in parallel for the determination of sTg levels.
This schedule could not be completed in all patients, i. e., 5 patients of the RITfirst group and 6 patients of the RITrep completed only 3 time points.
Fig. 1 Schematic representation of the assessment schedule.
A 3-level evaluation of the therapy response took place 3 months after RIT by means
of: a) whole-body scintigraphy (scintigraphic evidence); b) sTg levels (laboratory
evidence); and c) ultrasound of neck region and/or imaging of the affected region,
via low-dose computed tomography (CT) or magnetic resonance imaging (MRI) (morphological
evidence).
The preparation and count of CEC proceeded according to an established method [9 ]
[12 ]
[13 ]. One milliliter (ml) of whole blood in ethylene diamine tetraacetic acid (EDTA)
was subjected to erythrocyte red blood cell lysis using a solution of 100 mmol/l KHCO3 , 1.55 mol/l NH4 Cl, 1 mmol/l EDTA in distilled water. This suspension was centrifuged for 10 min at
1 200 rpm to separate the cell fraction from the plasma. The cell fraction, mostly
consisting of leucocytes but also containing CEC, was resuspended with 500 µl of PE
(PBS buffer [phosphate buffered saline] with 2 mmol/l EDTA) and 20 µl of this suspension
were exposed to a murine fluorescence-labeled antibody directed against EpCAM (EpCAM-fluoroisothiocyanate
[FITC]). In parallel, the cells were also stained with 7-amino-actinomycin D (7AAD),
a marker of cell viability, provided that only viable CEC were considered for further
analysis. After 30 min of incubation, the cell suspension was transferred to a well
of a 96-well-plate and investigated using a half-automated fluorescence microscope
(Scan_R, Olympus, Munich, Germany). 2 wells, and therefore 2 independent measurements
were performed for each patient. Cells were regarded as vital CEC when exhibiting
a strong EpCAM-FITC-signal and lacking accumulation of 7-AAD within the nucleus ([Fig. 2 ]). The mean value of the 2 measurements was used for further analysis. This method
allowed a reliable count of the number of CEC per milliliter of blood.
Fig. 2 Fluorescence image of a circulating epithelial cell (CEC) (#). The clear signal (green)
generated by the epithelial cell adhesion molecule (EpCAM) coupled to fluorescent
isothiocyanate (FITC) indicates that the cell is of epithelial origin. Cell fragments,
debris of lysed erythrocytes (*) and non-stained leucocytes (+) are also visible.
Instead of using absolute CEC values of individuals or their means within groups,
the therapy response was estimated on the basis of the CEC numbers before RIT (set
to 100%) and the percent change of CEC numbers at the chosen time-points post RIT
in comparison to baseline. This procedure based on relative changes allowed the comparability
of the time-course changes between individual patients, because CEC levels may differ
much between individuals [10 ]. Thus, the influence of outliers was minimized. In addition to the mean percent
changes of CEC within each group, the data were also analyzed as categorized pattern
of response (decrease or increase of CEC) within each group at each time point, in
order to depict the course of therapy response in individual patients. The same general
procedure was adopted for sTg levels.
Statistical analyses were performed using SPSS 19.0 (SPSS Inc., New York, NY, USA).
Group differences were tested by means of the Mann-Whitney U test. Correlation analyses
were performed by means of the Pearson correlation coefficient. The significance level
was set at p≤0.05.
Data were presented as means±standard deviation (SD).
Results
Absolute CEC numbers
Mean values and standard deviations of absolute CEC numbers of RITfirst and RITrep are shown in [Table 1 ] for each time point separately.
Table 1 Mean values±standard deviations of CEC/ml of RITfirst and RITrep at each time point.
RITfirst
RITrep
Number of CEC/ml
13 386±13 075
16 027±14 989
Number of CEC/ml
13 332±14 176
10 299±9 894
Number of CEC/ml
14 622±16 660
4 068±3 276
Number of CEC/ml
20 583±26 618
10 141±13 230
Clinical response
The mean age of the RITfirst group was 57.6±12.3 years. At the 3-month check post RIT, from a clinical point of
view, none of the patients showed signs of disease progression (scintigraphic, laboratory-wise,
or morphologic) ([Table 2 ]). One patient was lost to follow-up at the 3-month check (data not shown).
Table 2 Clinical response of the RITfirst and RITrep patients as assessed via sTg measurement, whole body scintigraphy and morphologic
imaging (i. e. sonography, CT or MRI). The diverging total number of patients is due
to a lost to follow-up.
RITfirst
RITrep
Disease Regression
Stable Disease
Disease Progression
Disease Regression
Stable Disease
Disease Progression
Laboratory (sTg levels)
7/12 (58%)
5/12 (42%)
0/12 (0%)
6/14 (43%)
6/14 (43%)
2/14 (14%)
Scintigraphic
9/12 (75%)
3/12 (25%)
0/12 (0%)
6/15 (40%)
7/15 (47%)
2/15 (13%)
Morphologic (imaging)
8/12 (67%)
4/12 (33%)
0/12 (0%)
5/15 (40%)
7/15 (47%)
2/15 (13%)
The mean age of the RITrep group was 61.6±15.2 years. At the 3-month check post RIT 2/15 patients (13%) scintigraphically
showed disease progression. In terms of sTg levels and morphologically also 2 patients
showed a progression ([Table 2 ]). In one patient, sTg measurement could not be performed.
CEC response
In terms of numbers of CEC, at day 2 approximately half of the patients receiving
a first treatment (RITfirst ) showed a reduction or an increase of CEC. This pattern of response remained similar
at day 14 and at 3 months ([Fig. 3 ]).
Fig. 3 Number of patients with changes of numbers of circulating epithelial cells (CEC)
and levels of serum thyroglobulin (sTg). Patients with differentiated thyroid carcinoma
(DTC) received a first radioiodine therapy post thyroidectomy (RITfirst ) or a repeated RIT (RITrep ) due to local persistence or metastatic recurrence of the DTC.
Differently from the RITfirst group, in the group of patients undergoing repeated RIT (RITrep ) more subjects showed decreased numbers of CEC at day 2 and 14 (9/14 [64%] and 9/13
[69%], respectively). 3 months after therapy, the proportion of patients with reduced
and increased CEC numbers was basically similar (7/13, 54% and 6/13, 46%, respectively),
at this time in analogy to the pattern observed at all time-points in the group of
patients undergoing a first RIT ([Fig. 3 ]).
CEC vs. sTg response
The sTg changes did not match the pattern of CEC changes. In the RITfirst group, for example, similar proportions of patients responded with an increase or
a decrease of sTg at day 2 and 14. At 3 months, all patients responded with a decrease
of sTg levels or stable conditions with sTg not detectable ([Fig. 3 ]).
In the RITrep group, in contrast, at day 2 post RIT nearly all patients (15/16, 94%) showed an
increase of sTg levels. At day 14, half of the patients (6/12, 50%) showed an increase
and half of the patients a decrease of sTg levels. This proportion (basically half
and half) remained evident at the 3-month time point ([Fig. 3 ]), similarly to the relative proportions of CEC cells at this time point.
Response patterns
The response to RIT was analyzed also in terms of percent change of CEC numbers and
sTg levels. In the RITfirst group, approximately half of the patients showed an increase of sTg levels to >250%
at day 2 or 14 (7/13 patients, 54%). In 5 of these 7 patients (71%), in contrast,
there was either a reduction of CEC to less than 60% or of at least 950% points after
an initial increase (see representative cases in [Fig. 4 ]).
Fig. 4 Time course of percent changes of CEC and sTg in the RITfirst group (mean±SD; upper left panel). On average, the time course was fairly parallel
for the 2 parameters, especially in consideration of the large variability of the
data. In contrast, the other panels show representative patients with clearly diverging
results at 1 of the 2 short-term time points (sTg increase to >250% of baseline levels,
CEC decrease to <60%, or, in the case of Patient 5.12 decrease of >950% points after
initial increase).
In the RITrep group, the mean sTg levels at day 2 and 14 were slightly elevated compared to baseline
(129±45% and 117±91%, respectively) ([Fig. 5 ], left panel). Clear increases of sTg levels to >250% were observed only in 2/16
patients (13%), whereas in these patients the CEC numbers decreased to <15% of the
baseline levels at 1 of the 2 early time points (see individual courses in [Fig. 5 ]).
Fig. 5 Time course of percent changes of CEC and sTg in the RITrep group (mean±SD) (upper left panel). On average, the time course was fairly parallel
for the 2 parameters, especially in consideration of the large variability of the
data. In contrast, the other panels show 2 patients with clearly diverging results
(sTg increase to >250% of baseline levels, CEC decrease to <15% of baseline levels
at least at 1 of the 2 short-term time points).
Correlation between clinical response/stage and CEC changes
Overall, neither of the 2 groups showed a significant correlation between the clinical
evidence of response (i. e., scintigraphic, laboratory, or morphologic) and the percent
change of CEC at any time point (data not shown). Furthermore, there was no clear
correlation between the initial TNM stage or number of involved lymph nodes, thyroid
remnant mass or radioiodine uptake on the one hand and the percent change of CEC on
the other hand at any time point (data not shown).
Sensitivity/Specificity/npv/ppv
To explore the potential of CEC changes as early predictors of therapy response, the
sensitivity (i. e., the correct identification of patients with clinical progression
by a CEC increase) and specificity (i. e., the correct identification of patients
with clinical regression by a CEC decrease) were also calculated ([Table 3 ]). The negative predictive value (npv), i. e., the ability of CEC decreases to predict
a disease regression or stability, for CEC changes at day 2 amounted to 90% when referring
to the scintigraphic results and to 100% when referring to morphologic and sTg response
([Table 3 ]).
Table 3 Sensitivity, specificity, and negative and positive predictive value of CEC changes
in the group of patients with repeated radioiodine therapy (RITrep ), with respect to different evidence of therapeutic response (serum thyroglobulin
(sTg); scintigraphic; morphologic). A clinical response was defined as regression
or stable conditions, progression as unambiguous disease aggravation.
Clinical assessment of therapy response at 3 months
Time point post RIT
Sensitivity (%)
Specificity (%)
Positive predictive value (ppv) (%)
Negative predictive value (npv) (%)
Laboratory (sTg levels)
day 2
100
67
20
100
day 14
0
67
0
75
3 months
50
55
17
86
Scintigraphic
day 2
50
67
20
89
day 14
0
73
0
89
3 months
100
64
33
100
Morphologic (imaging)
day 2
100
69
20
100
day 14
0
70
0
78
3 months
50
55
17
86
Correlation between activity and CEC changes
In the whole patient population there was no recognizable correlation between the
applied radiotherapy activity and the CEC changes at any of the time points investigated
(data not shown). In general, all patients showed a therapy response in terms of CEC
numbers, i. e., there were no cases in which CEC numbers remained equal to pre-treatment
levels.
Discussion
The present study evaluated the effects of I-131 RIT on the number of epithelial cells
circulating in the blood of patients with DTC. A specific focus was to monitor short-term
effects, to explore the possibility that CEC changes can be an early predictor of
therapy response.
Given the long half-life of I-131 (8.02 days), clinically measurable therapy effects
of RIT are expected only after a certain time, therefore studies on short-term effects
of RIT on the thyroid gland and/ or parameters are limited. A retrospective analysis
by Bernier et al. has shown that a first RIT was followed by increased sTg levels
5 days after therapy, and that the degree of the sTg increase positively correlated
with the later response to therapy [14 ]. The present study confirmed these results, i. e., a temporary increase of sTg was
observed in both study groups, in the RITfirst group more pronounced than in the RITrep group ([Fig. 4 ],[5 ] upper left panel).
CEC were identified in the blood of both patient groups and changes over time were
also observed. Because the CEC changes varied both in extent and direction of the
response (increases and decreases), a focus on mean group effects proved insufficient
to describe the therapy effects. Thus, patterns of response were primarily considered
(number of patients with CEC increase or decrease), a procedure successfully employed
in previous studies [13 ]
[15 ]
[16 ]. Also, the number of patients was too limited to provide statistically robust results,
therefore the value of the present study is to provide results useful for the design
and sample size calculation of future studies.
Contrary to the working hypothesis underlying this study, the course of CEC changes
did not parallel the changes of sTg levels, in fact the direction of the response
was basically opposite. At short-term time points post RIT (day 2 and 14), indeed,
more patients of the RITrep group showed a sTg increase in parallel to more patients showing a CEC decrease ([Fig. 3 ]). Some explanations are possible: The RIT may cause damage to CEC circulating in
the blood. Because in this study only vital cells were counted, potentially necrotic
cells as a result of RIT targeting were not included in the count. This may explain
the general result that most patients had reduced CEC number post RIT compared to
pre RIT. Assuming that at least part of the CEC derive from the thyroid [12 ] and therefore contain the protein thyroglobulin, higher quantities of this protein
can be released in the serum in consequence of cell damage or necrosis. Indeed, studies
on protein production at different de-differentiation stages of thyroid carcinoma
have shown that thyroglobulin can be detected in most differentiated carcinomas [17 ]. Thus, an increase of sTg in parallel to a decrease of CEC may reflect an effective
targeting of circulating thyroid cells by RIT. In analogy, Duffy et al. have stated
that necrosis and apoptosis of tumoral cells are the cause of initial increase of
tumor markers after chemotherapy [18 ].
As a matter of fact, only the patients undergoing repeated RIT showed this type of
response. In contrast, nearly all patients of the RITfirst group showed an early increase of sTg, but this was accompanied by similar proportions
of patients responding with a CEC increase or decrease ([Fig. 3 ]). In the RITfirst group, the frequently incomplete surgical ablation of the thyroid gland may result
in a reservoir of normal thyroid cells disseminating into the circulation. The number
of CEC measured in blood, therefore, may represent a balance between the therapeutic
targeting of circulating cells and the mobilization of cells from the residual thyroid
tissue ([Fig. 6 ]).
Fig. 6 Schematic representation of possible effects of RIT on blood CEC and on thyroid tissue,
the latter either as post thyroidectomy remnant (as for example in the RITfirst group) or as local persistence/distant recurrence (as in the RITrep group). 1) Baseline conditions: thyroid-derived CEC containing thyroglobulin (Tg)
(in red) and CEC of different tissue sources; 2) In response to RIT, the Tg-containing
CEC undergo cell damage or death and release Tg in circulation at measureable levels
(serum-Tg; sTg); 3) The RIT targets also normal thyroid tissue, where inflammatory
changes caused by treatment lead to overexpression of EpCAM and mobilization of cells
into the circulation. The CEC in blood, therefore, do not distinguish between EpCAM-positive
carcinomatous cells and inflammation-activated, but benign cells. The balance of these
2 sources may change inasmuch as a predominance of cell destruction causes a CEC decrease
and predominance of mobilization causes a CEC increase.
If the measured CEC stem from normal thyroid cells (in contrast to stemming from carcinoma
cells as hypothesized in the case of patients with repeated RIT), the question arises
as to why normal cells express high levels of EpCAM. Indeed, EpCAM does not only function
as adhesion molecule, but is also involved in differentiation, proliferation, cell
migration, and signal transduction processes. Accordingly, EpCAM is found also in
benign proliferative and inflammatory conditions [11 ]
[19 ] and, in fact, own investigations have detected CEC also in benign, therapy-requiring
thyroid diseases [12 ].
If this working hypothesis is correct, then CEC in blood may reflect not only malignant
thyroid cells but also normal, iodine-avid thyroid cells responding to RIT. The presence
of CEC at initial post-surgical stages has already been investigated in breast carcinoma
and benign breast diseases. In these conditions, mobilization of benign epithelial
cells into circulation in reaction to surgical trauma has been clearly shown, in parallel
to the release of long-lived malignant CEC [20 ]. In the present study, the patients of the RITfirst group had undergone thyroidectomy 4 to 6 weeks before the study, and own studies
have shown that the highest number of CEC are actually found in thyroid patients after
surgical intervention [12 ]. Patients with complete thyroidectomy have been found to have thyroid remnants in
at least 93% of the cases [21 ]. Taken together, the previous and present observations suggest that in the RITfirst group the CEC are of thyroid origin (at least in part), that these cells undergo
reactive changes in response to RIT, and their numbers in blood reflect a mixture
of mobilization and destruction of CEC.
However, this hypothesis clearly requires verification in future studies. Using the
applied method, it is impossible to conclude the origin of the CEC. They may derive
from the thyroid, from thyroid cancer tissue or from a different epithelium (benign
or malignant). In cancer patients, it is more likely that these cells represent malignant
cells; however, it cannot be proven in this setting. Therefore, it is necessary to
develop a method to unambiguously characterize these cells on a single-cell-basis,
for example via tissue-specific analyses of CEC and demonstration of their malignant
or non-malignant nature via carcinoma-specific markers. This could be achieved by
the analysis of the mRNA in these cells. Another possibility is to investigate whether
these cells contain the thyroid specific protein thyroglobulin using a fluorescence
labelled anti-Tg-antibody, as performed previously by Ringel et al. [22 ].
The hypothesis of CEC destruction and subsequent thyroglobulin release was corroborated
by the course of some individual cases ([Fig. 5 ]). 2 patients of the RITrep group, in fact, responded with a pronounced sTg increase in parallel to a clear reduction
of the CEC (the latter below 15% of the baseline levels), at least at 1 of the 2 early
time points. Similar individual courses were observed also in the RITfirst group, where 5 patients reacted with a clear sTg increase in parallel to a CEC decrease
([Fig. 4 ]). On the other hand, other patients displayed different patterns, therefore other
mechanisms are also conceivable.
Considering clinical effects, there were no statistically significant differences
between the CEC changes of patients with clinical progression or regression (data
not shown). However, in the RITrep group it could be shown that – of the patients with a CEC reduction at day 2 – at
least 89% showed a good clinical response to RIT (stable or regressing course of disease)
at the 3-month check (see negative predictive value in [Table 3 ]). A short-term CEC check 2 days after RIT may therefore be helpful in identifying
patients with a better chance of no disease progression if they show decreasing CEC
numbers. This hypothesis must be clearly verified in larger studies.
At the same time the specificity of the CEC changes at day 2 post RIT was 67, 67 and
69% (in relationship to scintigraphic, laboratory, and morphologic evidence, respectively),
thus definitely not all responders were identified by this method. Furthermore, the
positive predictive value (ppv) of CEC numbers must be assessed as insufficient, given
that the ppv did not exceed 33% at any time point. Thus, the measurement of CEC is
clearly not suitable for the prediction of disease progression.
Another limitation of the present study is the assessment of clinical response after
3 months because therapy effects of RIT might last longer than this interval. Recommendations
for evaluation of ablation success range from 3–6 months [3 ] to 6–12 months [23 ]. In the present study 3 months as follow-up interval were chose because therapy
strategies differed between patients after that time, thus potentially influencing
CEC response.
Conclusion
This study showed for the first time that the numbers of CEC in DTC undergo changes
in response to RIT. The data allow to formulate the hypothesis that the destruction
of CEC through RIT may induce a short-term release of thyroglobulin in the blood.
However, some patients showed patterns of response that were not compatible with this
hypothesis. Also, the sample size was too small to draw general conclusions, however
the data provide a valuable basis for the design of future studies. A clear relationship
between the clinical outcome and the CEC changes could not be found, but the data
suggest that early CEC decreases may help identifying patients more likely to respond
to RIT. The early changes of CEC may represent a useful addition to the standard diagnostic
checks performed at 3–6 months post RIT. Future studies in DTC are necessary to investigate
whether CEC are of thyroid origin and carcinomatous nature.
