Successful live cell harvest from bisected sentinel lymph nodes research report

www.elsevier.com/locate/jim

Journal of Immunological Methods 291 (2004) 71–78

Research paper

Successful live cell harvest from bisected sentinel lymph nodes

research report

Bruce Elliotta,b,c,*, Martin G. Cookb,c, R. Justin Johna, Barry W.E.M. Powella,Hardev Pandhaa, Angus G. Dalgleisha

aOncology Department, St. George’s Hospital Medical School, Cranmer Terrace, London SW17 0RE, United KingdombMelanoma Unit, St. George’s Hospital, Blackshaw Road, London SW17 0QT, United Kingdom

cDepartment of Histopathology, Royal Surrey County Hospital, Guildford, Surrey GU2 7XX, United Kingdom

Received 5 February 2004; received in revised form 26 April 2004; accepted 30 April 2004

Available online 15 June 2004

Abstract

Sentinel lymph nodes provide an excellent opportunity to study early immune responses to cancer. However, harvesting live

cells has not previously been possible, because it conflicts with the need to preserve tissue for histological interpretation. This

study used scrape cytology on 26 sentinel and 8 non-sentinel nodes, harvested from 17 stage I/II melanoma patients undergoing

sentinel node biopsy. Numbers of viable cells harvested before and after cryopreservation were measured and the effect on

subsequent histology assessed. The mean number of cells harvested from 26 sentinel nodes was 7.06� 106 (range 0.1–32.2),

with a mean viability of 99.5% (range 87–100, lower 95% CI 98.5%). Furthermore, counts and viabilities were well maintained

after cryopreservation. Flow cytometry confirmed CD3+, CD20+ and lineage-1� /HLA-DR+ subpopulations, consistent with

T-lymphocytes, B-lymphocytes and dendritic cells, respectively. Importantly, there was no discernible change in histological

detail and the proportion of positive sentinel nodes remained unchanged. This technique will allow more functional and

quantitative approaches to sentinel lymph node research.

D 2004 Elsevier B.V. All rights reserved.

Keywords: Sentinel lymph node biopsy; Tissue harvesting; Lymphocytes; Dendritic cells; Melanoma

1. Introduction

Sentinel node biopsy (SNB) was established as an

accurate staging method to identify lymphatic spread

0022-1759/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.jim.2004.04.025

* Corresponding author. Oncology Department, St. George’s

Hospital Medical School, Blackshaw Road, Cranmer Terrace,

London SW17 0RE, United Kingdom. Tel.: +44-7768-043531;

fax: +44-2087-250158.

E-mail address: [email protected] (B. Elliott).

of melanoma in 1991 (Morton et al., 1992). It has

since been shown to be the best prognostic indicator

yet for stage I/II disease (Gershenwald et al., 1999),

though the results of large clinical trials investigating

its therapeutic role are awaited.

Sentinel lymph nodes (SLNs) are also an important

resource for scientific research. They contain cells

which direct the immune response against antigens

found in the corresponding microbasin. Dendritic cells

(DCs) patrol the tissues to sample antigens before

B. Elliott et al. / Journal of Immunological Methods 291 (2004) 71–7872

passing through afferent lymphatics to the SLN. Once

mature they prime naı̈ve T-lymphocytes with this

antigenic information to trigger specific immune

responses and ultimately cell lysis. There is now good

evidence that altering specific immune responses

through cancer vaccines improves overall survival,

at least in the context of melanoma (Hseuh et al.,

2002). However, the immune processes within senti-

nel nodes are largely unexplored.

One factor hampering progress is the overriding

need to preserve the structure of the SLN to properly

identify the presence of metastasis. Traditional meth-

ods used to obtain live cells from lymph nodes cause

complete disruption of histological architecture.

Microstaging of the SLN requires careful systematic

assessment to gauge tumour volume and allow prog-

nostic stratification, and any loss of cytological or

histological detail compromises this process and

would undermine the accuracy of the result.

Recent work has shown that live tumour-draining

lymph node cells harvested by scrape cytology are

comparable to cells extracted by total nodal dissoci-

ation in terms of viable yields, phenotypic character-

istics, T-cell functionality and T-cell expansion factors

(Vuylsteke et al., 2002). We adapted this technique to

harvest live cells from freshly excised SLNs draining

primary cutaneous melanoma, and show that this is

possible without affecting the rate of positivity for

metastasis.

Table 1

Comparison of characteristics between scraped node cohort and

control group

Scraped nodes Control

Number of patients 17 17

Mean age (years) 51.7 56.4

Mean Breslow thickness (mm) 1.9 2.9

Positivity for melanoma (%) 24 24

2. Patients and methods

All 17 patients involved in this study were listed

for SNB under general anaesthesia, having undergone

prior primary cutaneous melanoma excision. The

indications for SNB were primary lesions greater than

1 mm Breslow thickness or those exhibiting ulceration

or regression histologically. All patients gave their

informed consent to take part in the study after the

nature and possible consequences had been fully

explained. In addition to cell harvesting from the

sentinel node, permission was also sought to remove

an additional ipsilateral non-sentinel node to provide a

control group. The study was approved by the Wands-

worth Local Ethics Committee.

To compare rates of positivity between SLNs un-

dergoing scrape harvest and those which did not, a

control group of 17 consecutive patients were selected

from a separate contemporaneous SNB study. The

characteristics of these two cohorts are detailed in

Table 1.

Excised lymph nodes were transferred into a 50-ml

flask containing 25 ml complete medium (RPMI 1640

with 2 mM L-glutamine, 100 units/ml penicillin, 0.1

mg/ml streptomycin (Sigma-Aldrich, Dorset, UK) and

10% fetal calf serum (Invitrogen, Paisley, UK)) and

taken to the laboratory for processing.

Excess fat was first trimmed from the periphery of

the lymph node, taking care to avoid a breach of the

capsule. The long axis was identified by rolling the

node between thumb and index finger to palpate its

dimensions. It was then carefully bisected along this

axis with a no. 20 scalpel blade (Swan-Morton, Shef-

field, UK) (Fig. 1a). The blade was gently drawn

across the freshly cut surfaces of the bivalved lymph

node, initially at an angle of about 40j. As the far

edge was reached, the angle was reduced and the

blade lifted flat in a slight scooping movement

(Fig. 1b). The residue left on the blade was then

agitated in a 15-ml flask containing 2 ml of complete

medium to release its cells into fluid (Fig. 1c), and the

process repeated two or three times per cut surface.

Finally, the cut surfaces were irrigated with a further 1

ml of complete medium to collect any loose cells not

lifted off by the scalpel blade. Before proceeding

further, the bisected lymph node was placed cut

surface down in a histology processing cassette and

immersed in a formalin container for histological

processing later.

The cell suspension was centrifuged at 500� g

for 5 min to aggregate the cells into a pellet. Cells

were resuspended in 2:1 complete medium/red cell

lysis buffer (Gentra Systems, Minneapolis, USA) for

5 min before washing as described above and cell

viability assessed by trypan blue dye exclusion. After

Fig. 1. Method of live cell harvest from sentinel lymph nodes. (A)

Using a broad scalpel blade, sentinel lymph nodes were initially

bisected along their long axis to leave two flattened hemispheres.

(B) The blade was then gently drawn across each freshly cut surface

at such an angle as to avoid cutting into the node; as the far edge

was reached, the blade was lifted flat in a careful scooping motion to

ensure collection of all harvested cells. (C) The resulting residue

was released from the blade into fluid by agitating in a 15-ml flask

containing 2 ml of complete medium. The process was then

repeated two or three times per cut surface.

Fig. 2. Comparison of cell harvest yields from sentinel and non-

sentinel nodes. The mean number of cells harvested from 26

sentinel nodes was 7.06� 106 (range 0.1–32.2), while the mean

number of cells harvested from eight non-sentinel nodes was not

significantly different (mean 12.4� 106, range 0.6–47.2), p= 0.39.

Error bars indicate standard error.

B. Elliott et al. / Journal of Immunological Methods 291 (2004) 71–78 73

one further wash cycle, cells were chilled on ice for

10 min before suspension in 1 ml per 107 cells

freezing medium (90% fetal calf serum, 10% di-

methyl sulfoxide (Sigma-Aldrich)) and frozen at � 1

jC/min to � 80 jC. After 48 h, the vial containing

the suspension was transferred into liquid nitrogen

for storage.

Thawed aliquots of cells were warmed rapidly

and washed in flow buffer (phosphate-buffered sa-

line with 0.1% sodium azide, 0.5% bovine serum

albumin (Sigma-Aldrich) and 1% human IgG

(Novartis Pharmaceuticals, Frimley, UK)) at room

temperature and cell viability assessed by trypan

blue dye exclusion. Cells were further washed in

flow buffer at 4 jC, and fluorochrome-conjugated

monoclonal antibodies added to the cell pellets (PE-

Cy5 CD3, HLA-DR PE and Lineage Cocktail 1

FITC, BD Biosciences, Oxford, UK) before incubat-

ing for 30 min on ice in the dark. The cells were

again washed in flow buffer before resuspension in

200Al Cell Fix (BD Biosciences). Flow cytometry

analysis was recorded at 10,000 events per measure-

ment for T-cells and 50,000 events for dendritic

cells, using the FACScan flow cytometer (Becton

Dickinson, Heidelberg, Germany).

All laboratory procedures were performed by the

same author (BE), and all histological interpretation

were performed by one histopathologist (MC).

2.1. Statistical analysis

The sentinel and non-sentinel node counts were of

Gaussian distribution (Kolmogorov and Smirnov

0.27 and 0.26, respectively, p>0.05) with differing

standard deviations (7.84 vs. 16.1) and so the Welch

corrected unpaired t-test was used for comparison.

The inguinal lymph node counts did not form a

Gaussian distribution (Kolmogorov and Smirnov

0.36, p= 0.02) and so comparison of counts obtained

from the three different lymphatic basins was carried

out using the Kruskal–Wallis test (nonparametric

ANOVA), with median values substituted for mean

values. Similarly the yields from the positive cohort

formed a non-Gaussian distribution and so the non-

parametric Mann–Whitney test was used to compare

these data with the yields from negative nodes.

p< 0.05 was considered significant.

3. Results

Live cell harvest proved feasible in all 34 nodes

attempted. It was possible to obtain a non-sentinel

Fig. 3. Cell harvest yield by lymphatic basin. The median number of

cells harvested from five head and neck lymph nodes was

significantly lower those from 12 axillary and 17 inguinal lymph

nodes (0.6� 106 vs. 5.75� 106 vs. 5.70� 106, respectively),

p< 0.01. Error bars indicate standard error.


control node without significant extra dissection in just

under half of the basins (8/17 or 47%). Time taken

from collecting the nodes in theatre to deposition in

the � 80 jC freezer was consistently under 90 min.

The mean number of cells harvested from 26

sentinel nodes was 7.06� 106 (range 0.1–32.2), with

a mean viability of 99.5% (range 87–100). The mean

number of cells harvested from eight non-sentinel

nodes was not significantly different (mean 12.4�106, range 0.6–47.2, p= 0.39) with a uniform 100%

viability (Fig. 2).

Of 34 nodes, five were taken from the head and

neck, 12 were taken from the axilla and 17 from the

groin. The median number of cells harvested from the

head and neck lymph nodes was significantly lower

than those from the axillary or inguinal basins

Fig. 4. Adequate numbers of live cells are recovered after cryopreservation

nitrogen, 17 were thawed and their cells counted. (A) The mean count wa

trypan blue dye exclusion was 93.6% (range 81–100). Direct comparison

divided into separate vials for cryopreservation, and not all vials were tha

therefore was less than 107 cells. Error bars indicate standard error.

(0.6� 106 vs. 5.75� 106 vs. 5.70� 106, respectively),

p < 0.01 (Fig. 3).

Of 39 sentinel and non-sentinel cryopreserved

cell aliquots, 17 were thawed and their cells

counted. The mean duration of cryopreservation

was 63 days. The mean count was 4.21�106 (range

1.0–8.9) with a mean viability of 93.6% (range 81–

100) (Fig. 4).

Flow cytometric analysis of six lymph nodes

revealed that the mean proportion of total cells

harvested expressing the T-lymphocyte marker

CD3 was 53.5%; a mean 40.6% expressed the B-

lymphocyte marker CD20, and a mean 1.75% of

cells were found to be lineage-1 cocktail negative/

HLA-DR positive in keeping with a dendritic cell

phenotype (Fig. 5a and b).

Four out of seventeen patients (24%) were later

found to have at least one positive sentinel node on

histological processing. This rate was matched by the

selected control group, in spite of a thinner mean

Breslow thickness in the scraped cohort (Table 1). The

mean number of cells harvested from the positive

sentinel nodes did not differ from the mean numbers

harvested from negative nodes (7.7� 106 vs. 8.6�106, respectively, p = 0.5).

The histopathologist reviewing the slides did not

detect any deterioration of histological or cytological

detail on specimens which had been scraped (Fig. 6).

The laboratory technicians also found no difficulty in

preparing slides from scraped specimens.

. Of 39 sentinel and non-sentinel cell aliquots cryopreserved in liquid

s 4.21�106 (range 1.0–8.9); (B) the mean viability as measured by

with fresh yields is not possible as harvests greater than 107 were

wed for analysis. The maximum possible yield for a thawed aliquot


4. Discussion

This study shows that harvesting live cells from

sentinel lymph nodes is feasible without affecting

Fig. 5. Flow cytometric analysis of recovered lymph node cells. The me

expressed surface markers CD3 (T-lymphocyte) and CD20 (B-lymphoc

dendritic cells were sought using dual-colour flow cytometry by staining fo

derived dendritic cells. (A) The mean expression of CD3 was 53.5%, the m

lineage-1 cocktail (negative) and HLA-DR (positive) was 1.75%. Error bars

are seen in the lineage-1 cocktail negative/HLA-DR positive cluster in th

described previously for blood-derived dendritic cells.

subsequent histological interpretation. Moreover, cell

yields and viability are sufficient for further analysis

by most cellular techniques. Quantities are such that

some compromise in technique should still provide

an proportion of total cells harvested from six lymph nodes which

yte) were measured by single-colour flow cytometry. In addition,

r lineage-1 cocktail and HLA-DR as previously described for blood-

ean expression of CD20 was 40.6% and the mean co-expression of

indicate standard error; (B) a well-circumscribed population of cells

e upper left quadrant, conforming with the dendritic cell phenotype

Fig. 6. Histological and cytological detail is preserved after scrape

cell harvest. No difference was noted on histological examination of

scraped as compared with unscraped sentinel lymph nodes. (A)

H&E stained section from a scraped sentinel lymph node showing

preservation of cytological detail; B) Pan Melanoma Plus (HMB45,

MART1 and tyrosinase) stained section from a scraped sentinel

lymph node showing a subcapsular cluster of melanoma cells,

confirming that immunohistochemical assessment after scraping is

satisfactory.


acceptable numbers of cells, and the relative lack of

attrition after cryopreservation in liquid nitrogen per-

mits later batching of samples.

The technique requires a degree of manual dexter-

ity which improves with experience. Initially, the

overriding principle was to avoid damaging lymph

node architecture; early scrapes were very gentle until

some confidence was built up by both harvester and

histologist. The difficulty increased with smaller

nodes and this was reflected in the lower yields

obtained from those nodes excised from the head

and neck. These are typically both greater in number

and physically smaller than those found in other

basins. Although no formal correlations were drawn,

smaller nodes appeared to offer lower yields.

Any information derived from live sentinel node

cells is clearly more valuable when considered in

conjunction with data from an adjacent non-sentinel

lymph node control. It was only possible to obtain a

control lymph node on just under half the operative

procedures for which it was sought. We obtained

consent for this additional procedure with the provi-

so that minimal extra dissection would be employed

to search for the additional node, so that the post-

operative risk profile would not change appreciably.

In particular, the removal of one additional lymph

node from a chain of 15 to 40 (according to basin)

was not deemed likely to increase the already low

risk of complications following SNB. This was

confirmed in a previously published paper from our

unit which found no significant difference in post

operative morbidity between patients divided accord-

ing to the number of SLNs excised per basin

(Hettiaratchy et al., 2000). In the event, all patients

gave their informed consent for excision of an extra

node, and the reasons for failure to obtain a non-

sentinel node were largely operative. Sentinel node

biopsy is designed to be less invasive than other

lymphadenectomy procedures and other lymph nodes

were not always encountered close to the sentinel

node.

Our justification for the adoption of this technique

hinges on its atraumatic effect on the lymph node.

Sentinel lymph nodes from our unit are processed

using a modified version of the technique originally

described by Cochran (1999), whereby 18 sections

are taken in six steps at 50 Am intervals from each

bisected surface (Cook et al., 2003). Cochran’s

technique stipulates a relative, rather than an abso-

lute, central plane for sampling and so it would seem

unlikely that the minimal tissue loss along the

bisection surface caused by scraping would materi-

ally affect sampling. The histopathologist did not

detect any loss of cytological detail or indeed any

difference as compared to over 1700 sentinel nodes

viewed during the preceding five years. Identical

rates of positivity between sentinel nodes subjected

to scrape and a control group not subjected to it

corroborate the lack of distortion of histological

features.


Scrape dissociation of lymph nodes has been used

for other applications previously. Shidam’s et al.

(1984) reported superior diagnostic accuracy from

249 scraped tissue samples compared to frozen sec-

tion or imprint cytology in the setting of intraoperative

tumour diagnosis. Dabbs et al. (1995) used the tech-

nique for rapid immunoperoxidase antibody staining

and advocated its use as an adjunct to frozen section.

Blumenfeld et al. (1998) compared the technique with

touch preparation and fine needle aspiration for diag-

nostic cytology of breast, lung and colon lesions. One

study looked at diagnostic scrape cytology of sentinel

nodes taken from 148 patients with breast cancer

(Smidt et al., 2002). However, no technical details

were offered about the procedure, the harvested cells

were fixed and there was no safety comparison made

with an appropriate non-scraped control group to

ensure that the procedure itself was not responsible

for false negativity. These authors were therefore

unable to answer the questions posed by the present

study.

Vuylsteke et al. (2002) used a similar harvesting

technique for a comparison with total nodal dissoci-

ation in tumour draining lymph nodes taken from

five patients undergoing regional lymphatic clear-

ance. They found that viabilities and phenotypic

characteristics of collected T-lymphocytes and den-

dritic cells did not differ significantly, and that T-

lymphocyte functionality and T-lymphocyte expan-

sion factors were also comparable. The crucial issue

of its effect on subsequent diagnostic processing was

not addressed, however, and this remains a pivotal

test before its acceptance for use in the sentinel

lymph node setting. Our paper shows that it is

possible to overcome these difficulties. It also further

validates the technique by contrasting viable yields

between sentinel and non-sentinel nodes, nodes from

different basins, and fresh and freeze-thawed sam-

ples. As far as we are aware, it also shows for the

first time that it is possible to detect dendritic cells

from lymph nodes by flow cytometry using lineage

cocktail negative/HLA-DR positive gating as previ-

ously shown from blood (Olweus et al., 1997).

We already know that individual tumour-draining

lymph nodes exhibit a marked variation in immune

function, with significant suppression found in those

closest to the primary melanoma lesion (Cochran et

al., 1989). Although sentinel nodes provide a far more

scientific model for investigation, most published

studies have used immunohistochemistry for analysis

(Huang et al., 2000; Cochran et al., 2001) and have

therefore lacked quantitative or functional informa-

tion. The ability to harvest live cells allows other

modalities to be employed such as flow cytometry,

cell sorting and functional studies.

The technique will be of considerable interest to

practicing sentinel node surgeons who have access to

laboratory facilities. The ability to cryopreserve live

cells from the sentinel node permits indefinite storage

of important immune information about each patient.

Such information may be useful when stratifying

patients for recruitment into cancer vaccine studies

in the future. Further work is required to compare

flow cytometry on live cells with subsequent paraffin-

embedded immunohistochemistry on the same nodes

in order to further validate the consistency of these

approaches. Thereafter the inner workings of the

sentinel node can be expected to provide many new

exciting targets for novel therapies in stage III and IV

malignancy.

Acknowledgements

The authors would like to thank all the surgeons

and theatre staff who helped in the collection of

intraoperative lymph nodes samples, and Margaret

Green for her help with slide preparation and

histology results. This research was funded by the

Cancer Vaccine Institute, SGHMS, London, SW17

0RE.

References

Blumenfeld, W.H., Hasmi, N., Sagerman, P., 1998. Comparison of

aspiration, touch and scrape preparations simultaneously

obtained from surgically excised specimens: effect of different

methods of smear preparation on interpretive cytologic features.

Acta Cytol. 42, 1414.

Cochran, A.J., 1999. Surgical pathology remains pivotal in the

evaluation of ‘‘sentinel’’ lymph nodes. Am. J. Surg. Pathol. 23

(6), 686.

Cochran, A.J., Wen, D.R., Farzad, Z., et al., 1989. Immunosuppres-

sion by melanoma cells as a factor in the generation of meta-

static disease. Anticancer Res. 9, 859.


Cochran, A.J., Morton, D.L., Stern, S., Lana, A., Essner, R., Wen,

D.R., 2001. Sentinel lymph nodes show profound downregu-

lation of antigen-presenting cells of the paracortex: implica-

tions for tumor biology and treatment. Mod. Pathol. 14 (6),

604.

Cook, M.G., Green, M.A., Anderson, B., et al., 2003. The devel-

opment of optimal pathological assessment of sentinel lymph

nodes for melanoma. J. Pathol. 200, 314.

Dabbs, D.J., Hafer, L., Abendroth, C.S., 1995. Intraoperative im-

munocytochemistry of cytologic scrape specimens. Acta Cytol.

39, 157.

Gershenwald, J.E., Thompson, W., Mansfield, P.F., et al., 1999.

Multi-institutional melanoma lymphatic mapping experience:

the prognostic value of sentinel lymph node status in 612 stage

I or II melanoma patients. J. Clin. Oncol. 17, 976.

Hettiaratchy, S.P., Kang, N., O’ Toole, G., Allan, R., Cook, M.G.,

Powell, B.W.E.M., 2000. Sentinel lymph node biopsy in malig-

nant melanoma: a series of 100 consecutive patients. Br. J. Plast.

Surg. 53, 559.

Hsueh, E., Essner, R., Foshag, L., et al., 2002. Prolonged survival

after complete resection of disseminated melanoma and active

immunotherapy with a therapeutic cancer vaccine. J. Clin.

Oncol. 20, 4549.

Huang, R.R., Wen, D.R., Guo, J., et al., 2000. Selective modulation

of paracortical dendritic cells and T-lymphocytes in breast can-

cer sentinel lymph nodes. Breast 6 (4), 225.

Morton, D.L., Wen, D.R., Wong, J.H., et al., 1992. Technical details

of intraoperative lymphatic mapping for early stage melanoma.

Arch. Surg. 127 (4), 392.

Olweus, J., BitMansour, A., Warnke, R., et al., 1997. Dendritic cell

ontogeny: a human dendritic cell lineage of myeloid origin.

Proc. Natl. Acad. Sci. 94 (23), 12551.

Shidam, V.B., Dravid, N.V., Grover, S., Kher, A.V., 1984. Role of

cytology in rapid intraoperative diagnosis: value and limitations.

Acta Cytol. 28, 477.

Smidt, M.L., Besseling, R., Wauters, C.A.P., Strobbe, L.J.A., 2002.

Intraoperative scrape cytology of the sentinel lymph node in

patients with breast cancer. Br. J. Surg. 89, 1290.

Vuylsteke, R.J., van Leeuwen, P.A., Meijer, S., et al., 2002. Sam-

pling tumor-draining lymph nodes for phenotypic and func-

tional analysis of dendritic cells and T cells. Am. J. Pathol.

161, 19.

Successful live cell harvest from bisected sentinel lymph nodes research report

Documents

Transcript of Successful live cell harvest from bisected sentinel lymph nodes research report