LN: lymph node; CTCL: cutaneous T cell lymphoma; MF:
mycosis fungoides; FC: flow cytometry; FNA: fine needle
aspiration; NCI-VA: National Cancer Institute–Veterans
Administration; ISCL/EORTC: International Society for Cutaneous
Lymphomas/European Organization for Research Treatment of
Cancer; TCR: T cell receptor; PCR: polymerase chain reaction;
SEM: standard error of the mean; ANOVA: analysis of variance
test; K-W: Kruskal-Wallis test; DSS: disease-specific survival
Enlargement of peripheral lymph nodes (LNs) is associated
with a poor prognosis in patients with mycosis fungoides (MF)
and Sézary syndrome (SS), the most common presentations of cutaneous T cell lymphoma (CTCL) [1,21]. LN enlargement
is usually the result of dermatopathic lymphadenopathy
(paracortical T zone hyperplasia with melanophages) combined
with variable degrees of infiltration by neoplastic T cells.
1Atypical/neoplastic cells defined as (1) small or large cells with irregularly
folded, convoluted nuclear outlines, a cerebriform appearance, and nuclear
hyperchromasia; (2) immunoblastic or transformed lymphoid cells with slightly
basophilic or clear cytoplasm and large nuclei with a vesicular chromatin pattern
and prominent nucleoli; and (3) large cells with multiple or multilobed nuclei and
cells closely resembling Reed-Sternberg cells.
For clinical staging, the ISCL/EORTC recommends that
peripheral LNs that are 1.5 cm or larger in diameter be excised
and examined for involvement by routine histopathology 2[3].
Ancillary studies such as molecular genetic studies to detect T
cell clonality (ideally the same clone as in involved skin) or flow
cytometry (FC) to detect lymphocytes with aberrant phenotypes
are helpful to confirm involvement and evaluate LNs that are not
overtly involved with lymphoma.
2B2 defined as absolute Sézary cell count ≥ 1.0/μL or CD4/CD8 ratio ≥ 10 plus
evidence of T cell clone or T cells with abnormal phenotype in the blood.
However, only peripheral LNs that are partially or completely
effaced by neoplastic T cells, i.e., histopathologic grades III/IV in
the Dutch scoring system [4] or the LN4 histopathologic grade
in the National Cancer Institute–Veterans Administration (NCIVA)
classification [5], are currently utilized for clinical staging.
This defines the N3 lymph node rating and perforce stage IVA2
(T1-4N3B0-2M0) [3]. Almost all effaced LNs have dominant T cell
clones that can be demonstrated by either Southern blot or PCRbased
methods [6-10]. Effaced LNs usually occur in the setting of clinically advanced CTCL and are associated with a significantly
worse prognosis compared to non-effaced LN patterns [2,11].
In the NCI-VA classification scheme, non-effaced LNs are
scored between LN0 to LN3 grades based on numbers of
atypical/neoplastic cells in parafollicular zones3: LN0: no atypical
cells; LN1: isolated atypical cells; LN2: small clusters of atypical
cells; and LN3: large clusters or sheets of atypical cells without
effacement of nodal architecture. The detection of T cell clones
in non-effaced LNs indentifies patients with a worse prognosis
compared to patients without a detectable clone [6-10,12], and
it has been suggested that LN3 grade nodes with evidence of a
clone, particularly when demonstrated with the Southern blot
method [6-8], should be considered comparable to LN4 nodes for
use in clinical staging. An N2b node rating is designated for such
cases, but currently it is not considered comparable to N3 rating
for staging of CTCL patients [3].
3Of interest, the latter patient’s disease subsequently evolved into SS, but at the
time of the initial LN biopsy, the absolute Sézary cell count was 113 K/μL, the
CD4/CD8 ratio was 66%/10% and Southern blot analysis showed no evidence
of a clone in the blood.
We and others reported that fine needle aspiration (FNA)
with evaluation of the sample by cytopathology, PCR analysis
of the T cell receptor (TCR) gene and FC provides an additional
means to assess LNs from patients with CTCL [13-15]. Specifically,
we observed a strong correlation between the grade I to IV
cytopathologic score based on numbers of atypical cerebriform
lymphocytes obtained by FNA and the NCI-VA score for 11 excised
nodal specimens [13]. However, although involvement of LNs
by CTCL (or other lymphomas) can be established by FNA, the
method does not reliably determine the degree of effacement of
nodal architecture which is a prerequisite for clinical staging, i.e.
N3 rating. The purpose of this study is to review our experience
with FC obtained on LNs from patients with MF/SS and determine
if a criterion based on FC for N3 node ratings can be established
for use with FNA samples.
FC was performed on a portion of 78 excised peripheral LN
from 72 patients with CTCL (32 diagnosed with MF and 40 with
erythrodermic CTCL (E-CTCL) which includes 21 patients with SS
(B2 blood rating4) and 18 with erythrodermic MF (E-MF) as defined
by ISCL/EORTC criteria [3]. The number of Sézary cells per 100
lymphocytes on blood smears was visually counted by one experienced
technician as previously reported [16], and the absolute Sézary cell
count calculated from absolute lymphocyte count determined from
the concurrent WBC and leukocyte differential count. The magnitude
of LN involvement (LN grade) was assessed using the NCI-VA
classification scheme [5]. In addition, the neoplastic cells of overtly
involved LNs were classified according to small, mixed small-large
or large cell morphology as previously described [11].
4LNs with neoplastic cells that did not express CD4 (one LN4, one LN0-2), coexpressed
CD4 and CD8 (one LN4 node) or were not studied for CD7 (one LN4 node) are excluded.
Two LN specimens (pathologic diagnosis in parentheses) were
obtained from one patient with MF and large cell transformation
in the skin (both dermatopathic lymphadenopathy), 2 patients
with SS whose disease progressively worsened despite treatment (dermatopathic lymphadenopathy and CTCL for one
patient and small cell and mixed small-large cell lymphoma
pattern for the other patient), and 1 patient with E-MF that
progressed to SS (both dermatopathic lymphadenopathy). Three
LN specimens were acquired from a patient with plaque phase
MF (focal involvement with MF) who progressed to the tumor
phase (dermatopathic lymphadenopathy), and subsequently
leukemic involvement (diffuse involvement with lymphoma). For
prognostic correlations, only LNs that were obtained at the time
of initial staging of 52 patients were used.
Single color flow immunophenotyping was performed on
52 excised LNs studied before 1997 as previously described5
[17]. The antibodies in the panel reacted against T-cell markers
(CD2, CD3, CD4, CD5, CD7, CD8), B-cell markers (CD19, CD20,
kappa and lambda light chains), natural killer (NK) cell markers
(CD16, CD57), activation/proliferation markers (HLA-DR, CD25,
CD71) and CD10. The ratios of CD4/CD8 and kappa/lambda were
provided by the laboratory. For 26 excised LN specimens studied
after 1997 and specimens obtained by FNA, two or three color
immunophenotyping that used CD45 and side scatter to define
the lymphocyte gate provided measurement of the CD4+CD7- and
CD4+CD26- lymphocytes as well as expression of CD16+CD56+
NK cells. FNA was also directly obtained on 4 excised LNs
for comparison with the portion submitted for FC; the latter
specimen was used for analysis [13]. The ratios of CD7/CD4 and
(CD7-CD8)/CD4 were calculated to estimate the expression level
of CD7 on CD4+ cells. Although both parameters correlated well
with available direct measurements of CD4+CD7- cells (rho=
-0.952 and rho= -0.948, respectively, P= 0.01), CD7/CD4 ratio
was chosen for comparative analysis. Southern blot analysis
of the TCR beta chain or PCR analysis of the TCR gamma chain
for evidence of a T-cell clone was performed on 45 excised LNs
(40 of the initial LNs). The presence of either cells expressing an
abnormal immunophenotype or a T cell clone was defined as a
positive ancillary study.
5We assume that LNs with histopathologic evidence of CTCL would be abnormal
by cytopathology and/or special studies. However, we acknowledge that
focal involvement might be missed by FNA.
FNA of enlarged peripheral LNs was performed directly on 18
patients and submitted for cytopathology and ancillary studies.
The clinical diagnosis was MF for 6 patients (5 at tumor phase, 1
at patch phase) and E-CTCL for 12 patients (9 with SS, 3 E-MF).
The cell yield was inadequate for one sample from a patient
with SS and sufficient to allow measurement of CD4+CD7- and
CD4+CD26- lymphocytes on 17 and 16 specimens, respectively.
CD19+ cells with κ/λ ratio were measured on 17 and 14 samples,
respectively. CD2, CD5, CD25 and HLA-DR were measured on 8
samples and will not be discussed further. PCR for T cell clonality
was performed on 16 of 18 cases.
Results of laboratory studies were given as mean values ±1
standard error of the mean (SEM) and/or median value with a
range. Fisher’s and Pearson’s chi-square exact tests were used to
test categorical data. Welch’s t-test, which does not assume equal
variances, was used to compare mean values of two independent samples. For 3 groups, mean values were compared using oneway
analysis of variance (ANOVA) together with the Games-
Howell post hoc test when differences were significant. The
nonparametric Kruskal-Wallis (K-W) test was also used to test for
differences in median values. Spearman’s correlation coefficient
was used to test for significant correlations. The Cox Proportional
Hazards model and Kaplan–Meier estimates of survivor function
were used for survival analysis. Deaths attributed to CTCL or its
treatment defined disease-specific survival (DSS) in these models.
The statistic -2 log L was used to compare alternative Cox models
(the lower the value of -2 log L, the better the fit in the model)
[18]. Statistical software used in the study were SYSTAT10 and
SPSS 13.0 for Windows, SPSS, Inc. (Chicago, IL) and StatXact-3,
Cytel, Inc. (Cambridge, MA).
The relationship between histopathologic diagnosis and clinical
diagnosis at the time of biopsy is given in Supplemental Table 1.
Of 78 excised LNs, 33 (42%) nodes had histopathologic evidence of
CTCL and the remaining nodes showed reactive changes, typically
dermatopathic lymphadenopathy. Involvement was more likely to
be present in LNs from patients with SS (17/24 nodes = 71%) than
patients with MF (14/35 nodes = 40%) or E-MF (2/18 nodes = 11%,
P< 0.001). The morphologic pattern of neoplastic cells in involved
LNs of SS and MF was not significantly different (P= 0.426).
NCI-VA histopathologic grades and results of FC and molecular
genetics were recorded for 72 excised LNs obtained for clinical
staging (Table 1 and Supplemental Table 2). Of 30 LNs diagnosed
with CTCL, 6 had paracortical expansion with neoplastic cells while
retaining germinal centers (LN3) and 24 had partial or complete
effacement of nodal architecture (LN4). Of the 42 non-involved LNs,
7 were graded as LN3 and none were LN4. Thus, LN3 grade was
recorded in 20% (6/30) and 21% (7/34) of LNs diagnosed as involved
or not involved with CTCL, respectively.
The prognostic implications of histopathologic grading of initial
LNs are shown in Table 2. As expected, histopathologic involvement signified a worse prognosis than non-involvement. The DSS curves
associated with LN0-1 and LN2 nodes were nearly identical and
therefore were combined for subsequent analysis (LN0-2). Overall,
the mean DSS of patients with LN0-2 nodes (14.42 years) was
significantly better than LN3 nodes (7.08 years; P= 0.026) and
LN3 nodes were better than LN4 nodes (2.32 years; P= 0.008,
Supplemental Figure 1). Of note, the difference in DSS for patients
with involved LNs classified by morphologic appearance was not
significant (P= 0.523).
Finally, when tested in the Cox model with age at biopsy entered
as a covariate, both clinical diagnosis and presence of histopathologic
involvement were significantly associated with deaths from CTCL
as the endpoint (-2 log L changed from 285.89 to 245.23). If LN
grades (LN0-2, LN3 and LN4) were substituted for histopathologic
involvement in the model, the fit with outcome improved only slightly
(-2 log L, 244.45).
Impact of Ancillary Studies
Of the 47 excised initial LNs studied by single antibodies,
13 (28%) had cells that had a diminished expression of CD7
consistent with an abnormal T cell population. One specimen
had in addition loss of CD2, one had loss of CD3 and CD4 and one
had co-expression of CD4 and CD8. The remaining 34 LNs were
considered to have a normal phenotype. However, 12 of these
specimens had CD4/CD8 ratios ≥ 10 (range, 10.8 to 42.7) and 2
expressed CD10.
Of the 25 LNs studied with combinations of antibodies, 16
(64%) had cells with an abnormal phenotype. Diminished CD7
expression occurred in 9 specimens as the only abnormality or
was combined with loss of CD4, CD5 or co-expression with CD10
(one case each). Four of the remaining LNs had cells expressing
a CD3dim phenotype (two also loss of CD26), one had CD2- cells,
one had CD3-CD30+ cells and one with CD7dimCD26- cells.
Altogether, of initial LNs scored for LN grades, cells with an
abnormal phenotype were detected by FC in 5 of 27 (19%) LN0-2 nodes, 6 of 13 (46%) LN3 nodes and 19 of 24 (79%) LN4 nodes
(1 of 8 LNs not scored). In the Cox model with age and LN grades
as covariates, evidence of cells with an abnormal phenotype was
not a significant variable.
Of the 43 LNs studied for clonality, a T cell clone was detected
by Southern blot analysis in 7 initial LNs and by PCR in 19 nodes.
The presence of a clone signified a significantly worse prognosis
(Mean DSS, 5.90 years ± 0.84, 5-year 57%) compared to patients
without a clone (16.55 years ± 2.28, 5-year 71%, P= 0.014). In
terms of LN grades, a clone was present in 7 of 20 (35%) LN0-2
nodes, 7 of 8 (88%) LN3 nodes and 10 of 12 (83%) LN4 nodes.
Although the presence of a clone on 7 LN0-2 nodes signified a
worse prognosis (Mean DDS, 7.61 years ± 1.62) compared to 13
LNs without a detectable clone (Mean DDS, 17.51 years ± 2.54),
the difference in survival curves was not statistically significant
(P= 0.163). In the Cox model with age and LN grades as covariates,
the presence of clonality was not a significant variable.
Altogether a positive ancillary study occurred in 43 of 72
(60%) initial LNs overall (Table 2). The DSS for patients with a
positive ancillary study (mean 4.99 years ± 0.68) was significantly
shorter than patients with a negative special study (mean 11.85
years ± 1.86; P= 0.010). However, in the Cox model with age
and LN grades as covariates, a positive ancillary study was not a
significant variable.
It should be noted that of 29 patients with a negative ancillary
study, 5 were diagnosed to have LN involvement by histopathology
(LN3 grade: 2 patients; LN4 grade: 3 patients). Two patients
with effaced LN4 nodes had SS. Both had absolute Sézary cell
counts exceeding 1.0 K/μL and T cell clone in the blood shown
by molecular genetic and chromosome analysis. The LN of one
patient showed a small cell pattern with decreased expression of
CD7 by immunohistochemistry even though CD7 was expressed
on the majority of cells studied by FC (74% CD4+, 70% CD7+). The
LN from the other patient with SS showed a mixed small-large cell pattern with patchy positivity for CD7 by immunohistochemistry
and 98% CD4+, 72% CD7+ by FC. Subsequent investigations
showed that the neoplastic cells of this patient also expressed
FoxP3 [19]. The third patient had diffuse effacement of the LN
with large cells and aneuploidy demonstrated by cytophotometry,
results consistent with transformed MF. FC showed 53% CD4+
and 53% CD7+ cells. Immunohistochemistry and molecular
genetic analysis were not performed. The two patients with
LN3 nodes had focal subcapsular involvement of the LNs, one
with large cells that expressed a CD4-CD8-CD7- phenotype by
immunohistochemistry, the other with a mixed small-large
pattern not further characterized by immunohistochemistry. It is
therefore likely that an ancillary study might have been positive
in some of these patients if modern FC and molecular genetic
methods that are more sensitive to detect small populations of
neoplastic cells than methods used in this study.
Prognostic Indicators Other than Lymph Node Histopathology
The results of FC according to LN grade of initial LNs are
shown in Supplemental Table 2. Several findings on FC correlated
significantly with prognosis in the Cox model with patients’ age
as the only covariate. These included a high CD4/CD8 ratio (P=
0.020) due mainly to low CD8+ cells (P= 0.020) rather than high
CD4+ cells (P= 0.119). When categorized as CD4/CD8 ≥ 10, the
level of significance changed to P= 0.003 and this remained
significant in the model with LN grade added as a covariate.
In addition, given that CD7 is often lost by neoplastic T cells,
[17,20-22] it was not surprising that low percentage of CD7+
cells (P= 0.002) and calculated CD7/CD4 ratio (P= 0.004) for all
patients and direct measurement of CD4+CD7- cells in a smaller
cohort (P= 0.025) also had prognostic importance with patients’
age in the model. However, these parameters no longer retained
significance with age and LN grade as covariates.
Also, the percentage of cells expressing the transferrin
receptor CD71 was significant (P= 0.003) as were the calculated
ratios of CD71/CD3 (P= 0.044) and CD71/CD5 (P= 0.017); CD71/
CD4 ratio was nearly significant (P= 0.084). Of interest, other
activation markers CD25 (P= 0.857) nor HLA-DR (P= 0.486) were
not significantly associated with survival. Again, CD71 no longer
was significantly associated with survival with age and LN grades
as covariates.
Lastly, we reasoned that the calculated CD4/CD19 ratio might
reflect the degree of nodal effacement and indeed this parameter
was significant with patients’ age in the Cox model (P= 0.047) but
not with LN grade (P= 0.941).
Possible Surrogates of LN4 Histopathologic Grade Based on Flow Data from Excised Lymph Nodes
We first plotted various flow parameters significantly
associated with prognosis from all 54 excised LNs graded as
LN0-2, LN3 and LN4 with the intent of identifying thresholds
that might differentiate LN4 from LN0-3 nodes. Considerable
overlap with LN3 nodes was observed for each parameter such
that potential diagnostic thresholds for LN4 define only a small
proportion of LN4 nodes. For example, a threshold for CD4/CD8
set at ≥ 30 identified 7 of 24 LN4 nodes, 2 of 15 LN3 nodes and none of 28 LN0-2 nodes (Supplemental Figure 2). This criterion
would correctly classify only 29% of LN4 node and misclassify
5% of LN0-3 nodes. Somewhat better results were achieved with
thresholds using CD7+ cells < 15% (9 of 25 LN4 or 36% positive
versus 1 of 43 LN0-3 or 2% misclassified) and CD4/CD19 ratio
≥ 10 (6 of 25 LN4 or 24% positive versus none of 43 LN0-3
misclassified; Supplemental Figures 3 and 4).
An alternative strategy was to define flow criteria that have
mean DSS rates similar to that of 24 LN4 nodes, i.e., within the
95% confidence interval for mean DSS of LN4 nodes (1.32 to 3.32
years). Only initial LNs that might be expected to show evidence
of neoplastic cells by FNA were included in this analysis. An
example of this approach for CD4/CD8 ratio as a criterion is
shown in Figure 1 and Supplemental Table 3. As the CD4/CD8
ratio is increased from ≥ 10 to ≥ 30, the estimated mean DSS for
patients becomes shorter, and at CD4/CD8 ≥ 30, the mean DSS
of 2.62 years was within the 95% confidence interval for LN4
nodes. Also, the 5-year DSS rate (22%) approximated that for
LN4 nodes (17%). However, although none of the LN0-2 nodes
had a CD4/CD8 ≥ 30, only 29% of LN4 nodes were positive for
this criterion. In a similar way, other flow criteria were evaluated
as potential surrogates for LN4 (Table 3). As with CD4/CD8 ratio,
these criteria identified only a minority of LN4 nodes.
Another possibility involves measuring the percentage of
CD4+CD7- and CD4+CD26- lymphocytes in LNs similar to what has
been proposed for the blood in CTCL. [20-25] Because CD7 and to a lesser extent CD26 may be variably expressed by neoplastic cells, it
has been proposed that both CD4+CD7- or CD4+CD26- populations
be measured and that the maximum percentage of either one (herein
called maxiCD4+CD7-/26- percentage) be used to estimate neoplastic
cell numbers in the blood. [26,27] Of the excised LNs in this series,
CD4+CD7- and CD4+CD26- were measured together for only 6
graded LNs. The 2 LN0-2 nodes had maxiCD4+CD7-/26- percentage
< 30% whereas both LN3 and 1 of the 2 LN4 nodes had values exceeding 40%. The exception was a LN diffusely involved with a
transformed CD3-CD4+CD30+ large cell lymphoma that expressed
both CD7 and CD26 and an identical T cell clone that was detected in
erythrodermic skin of a patient with E-MF.
Results of Fine Needle Aspiration
FNA showed cytopathologic evidence of neoplastic cells in
12 of 17 LNs that had adequate sampling. One additional LN
from a patient with tumor phase MF without abnormal cells by
cytopathology nor FC had evidence of a T cell clone that was
also in the skin and blood; this specimen was categorized to be
minimally involved with disease. The results of FC for 13 LNs
with and 4 LNs without evidence of neoplastic involvement by
FNA are shown in Table 4. Of interest, 3 patients with typical SS
had no detectable T cell clone in the LN specimen by PCR despite
having abnormal cells by cytopathology and CD4/CD8 ≥ 10 and
high percentages of CD4+CD26- cells (58%, 79% and 93%) by FC.
In their concurrent blood sample, a clone was also not detected by
PCR despite all demonstrating high Sézary cell counts, CD4/CD8
≥ 10 and high percentage of CD4+CD26- cells. A chromosomallyabnormal
clone was demonstrated in 2 patients. Therefore, it
seems likely that these patients are examples of “false negative”
PCR results that we encountered using PCR methodology at the
time in more than 20% of Sézary patients with a chromosomally
abnormal clone [27].
A box plot of CD4+CD7- and CD4+CD26- percentages
according to absence or presence of detectable neoplastic cells by
FNA is shown in Figure 2. For involved LNs, the percentage of
CD4+CD26- cells (median, 57.5%) was higher than CD4+CD7- cells
(median, 28.0%), but the difference was not statistically significant
(P= 0.142). CD4+CD26- percentage was higher than CD4+CD7- in
all but one case where CD4+CD7- cells were only 0.8% higher than
CD4+CD26-. Consequently, the percentage of maxiCD4+CD7-/26-
cells and CD4+CD26- cells was nearly identical and ranged between
1.5 to 93% (Table 4). All were above 20% except for a LN considered
to be involved solely on the basis of a detectable T cell clone by
PCR. Conversely, the 4 LNs without detectable neoplastic cells had
maxiCD4+CD7-/26- percentages ranging from 5 to 25%.
In terms of prognosis, the 13 patients with evidence of neoplastic
cells in their LNs by FNA had a shorter mean DSS (5.40 years ±1.43) compared to the 4 patients without evidence of neoplastic cells (8.94
years ± 2.27); however, the difference was not statistically significant
(P= 0.369). In the Cox model, for patients with LN involvement,
maxi-CD4+CD7-/26- percentages provided a significant but only
slightly better fit with disease-specific death as the endpoint than
CD4+CD7- or CD4+CD26- alone whereas CD4/CD8 and CD4/
CD19 ratios were not significant. The mean DSS and 5-year survival
rates for patients with maxi-CD4+CD7-/26- percentages in LNs ≥
30%, ≥ 40% and ≥ 50% were 3.97 years/22%, 2.92 years/13% and
2.74 years/14%, respectively.
Comparison of Fine Needle Aspiration with Excised Lns for Staging
The results of flow studies for involved LNs studied by FNA
and involved LNs that were excised at the time of initial staging
are compared in Supplemental Table 4. Altogether 13 of 17
(76%) LNs studied by FNA had evidence of neoplastic cells in
the sample compared to 36 of 52 (69%) LNs excised at the time
of initial staging. However, higher percentages of CD3+CD4+ T
cells and lower CD19+ B cells were observed in FNA samples
compared to excised LNs. Consequently, the calculated CD4/
CD19 ratio was also significantly higher in FNA samples (P<
0.001). These findings suggest that the LNs studied by FNA may
have had a greater magnitude of involvement, i.e., were more
effaced, than excised LNs. Although it is possible that FNA may
have disproportionally sampled the paracortical zones of LNs
where T cells predominant, the CD4/CD8 ratio and percentage
of CD4+CD7- cells in FNA samples was only slightly higher than
excised LNs.
In this study, we attempted to identify flow parameters that
might be used with FNA as surrogates for the LN4 histopathologic
grade that is currently used for clinical staging of CTCL [3].
Potentially this goal might be attained by counting number of
atypical cells per field on cytopathologic preparations [13],
quantifying the number of clonal T cells by molecular methods
[28], or determining percentage of lymphocytes with an abnormal
immunophenotype by FC. The problem with counting atypical
cells on cytopathologic preparations is that stimulated normal
lymphocytes can develop hyperconvoluted “cerebriform” nuclei, thereby making distinction between neoplastic versus normal
lymphocytes based on cell morphology alone difficult [29].
Using outdated single parameter FC on excised LNs, we
identified several criteria that had a similar DSS as LN4 nodes
obtained at the time of clinical staging (shown in Table 3). Three
of these criteria (CD4/CD8 ≥ 30, CD7+ cells ≤ 20% and CD4/
CD19 ≥ 8) were not positive in LNs graded as LN0-2, suggesting
a possible role for diagnosis of involvement. However, only a
minority of LN4 nodes fulfilled these criteria, indicating that a
different approach was required for staging.
In patients with CTCL, blood tumor burden has been
estimated by the percentage of CD4+CD7- or CD4+CD26- cells
[20-25]. Given that CD7 and CD26 expression by neoplastic T
cells in a given case may be variable (some cells positive, some
cells negative), it has been suggested that both populations be
measured, and the maximum value used for determining blood
tumor burden [26,27]. We now propose that a similar approach
be applied to LNs studied by FNA.
In this series, CD4+CD7- and CD4+CD26- lymphocytes were
simultaneously measured in 18 LNs with evidence of neoplastic
cells (6 LNs by excision, 12 LNs by FNA). The percentage of
CD4+CD26- cells was higher than the percentage of CD4+CD7- for
almost all cases. This is consistent with the observation that CD7
and CD26 expression by neoplastic T cells is lost by about 60%
and 90% of patients with SS, respectively [22]. Consequently,
the percentage of CD4+CD26- cells usually defines percentage of
maxiCD4+CD7-/26- cells.
Our data also suggest that patients with LNs that have percentages
of maxiCD4+CD7-/26- ≥ 40% have a prognosis comparable to patients with effaced LN4 nodes and this might be used with FNA as
a criterion for N3 node rating for clinical staging. However, we also
observed that maxiCD4+CD7-/26- ≥ 40% occurred in 3 of 5 excised
dermatopathic LNs without apparent histopathologic involvement.
The CD4+CD26- percentages were 45%, 49% and 60% whereas
the corresponding CD4+CD7- values were 14%, 23% and 27%.
Accordingly, we propose that the threshold for maxiCD4+CD7-/26-
as a surrogate for LN4 be increased to ≥ 50% plus evidence of
involvement to minimize “false positives” until additional cases have
been studied. The one patient with 60% CD4+CD26- cells had an
LN3 node with high CD4/CD8 ratio (85%/5%).
Of course, this proposed flow criterion might change as
more LNs are studied. One confounding factor with measuring
maxiCD4+CD7-/26- is that CD4 may not be expressed (CD4-CD8-
or CD4-CD8+) by neoplastic T cells as occurred in 3 patients in
this series or neoplastic cells might co-express CD7 and CD26 as
occurred for one patient. Other markers of neoplastic cells such as
antibodies directed against TCR-Vβ [30,31], Kir3DL2/CD158k
[32,33] or CD164 [34] need to be investigated.
The high percentage of CD71+ cells in involved LNs (median,
8.5%, range 1 to 56%) is consistent with a high proliferation
rate of neoplastic cells. [35] However, high values of CD71
might be encountered in non-neoplastic hyperplastic LNs [36],
and one of our LNs with follicular hyperplasia had 25% CD71+
cells plus evidence of a T cell clone. In future studies, it might be
worth examining the diagnostic and prognostic implications of
CD71 expressed on specific lymphocyte subsets, e.g., CD4+CD26-CD71+.
Finally, in accordance with other studies [2,37,38], evidence of clear-cut histopathologic involvement in LNs was frequent in patients
with advanced MF and SS and, if present at the time of staging,
signifies a worse DSS compared to non-involved LNs for patients
with MF and E-MF. However, for our patients diagnosed with SS,
overt LN involvement was not associated with a significantly worse
DSS compared to Sézary patients without involvement. We also
did not find a significant difference in survival for nodal infiltrates
composed predominately of small or large neoplastic cells as reported
previously [11]. These findings suggest that pathologic assessment of
LNs at the time of clinical staging may not be necessary for patients
diagnosed as SS.
The use of FNA to assess LNs in CTCL is controversial. One
concern is that neoplastic involvement might be focal and
thereby not be detected as compared to excision. The presence
of nodal involvement often determines the approach to therapy
for patients with MF, i.e., skin-directed only treatment versus
skin treatment combined with systemic agent [39]. However, for
clinical staging, the issue is also about whether the involved LN
is partially or completely effaced (LN4 grade). We believe FNA
combined with ancillary studies would identify involvement in
such LNs.
In conclusion, our data suggest that FNA coupled with
measurement of CD4+CD7- and CD4+CD26- lymphocyte subsets
and molecular analysis of T cell clonality might provide an alternative
means to assess enlarged peripheral LNs in patients with CTCL,
thereby decreasing the morbidity associated with excisional biopsies.
A possible approach might be to initially perform FNA first and, if
this is completely normal, then obtain an excisional biopsy if the
clinical suspicion of nodal involvement is high. Prospective studies
are required to determine the utility of this approach.
Dr. Vonderheid organized the data, performed the statistical
analysis and helped write the manuscript.
Dr. Hou supervised some of the flow cytometry and helped
write the manuscript.
Dr. Bigler supervised some of the flow cytometry and helped
write the manuscript.