author reply

16
CORRESPONDENCE Bone Marrow Staging of Patients with Non- Hodgkin’s Lymphoma Is Flow Cytometry a Useful Test? R epresenting a busy flow cytometry laboratory with an annual workload of approximately 1400 acute leukemia and malignant lymphoma cases, we read the article by Naughton et al. 1 with con- siderable interest. The authors concluded that “flow cytometry of bone marrow aspirates has a limited role in the routine staging and follow-up of patients with an established diagnosis of [non-Hodgkin’s lymphoma] NHL.” 1 However, this conclusion must be challenged because a number of methodologic flaws and inadequacies in the process, analysis, and interpretation are evident in this study. 2,3 The authors used Ficoll-Paque (Pharmacia Biotech, Piscataway, NJ) processed specimens rather than recommended erythrocyte ly- sis. 2 Ficoll-Paque centrifugation of bone marrow aspirates is known to result in the selective loss of some cell populations. This phenomenon is related to the fact that neoplastic lymphoid cells may have a different buoyant density from normal lymphocytes and may not be found where anticipated in the gradient. 2 The authors, possibly un- knowingly, thus may have reduced the number of potentially analyz- able malignant lymphoid cells in their samples, which may explain a portion of their staged cases with a negative flow result. Single color flow cytometry was used on 81 specimens and two color flow cytometry on the remaining 192 specimens. It currently is widely accepted that even two color flow cytometric analysis is inad- equate for an appropriate characterization of hematologic neoplasia and virtually precludes an accurate assessment of minimal residual disease. Five parameter analysis including forward and side scatters accompanied by at least three color analysis should constitute the minimal standard for immunophenotypic analysis of lymphoprolif- erative diseases, with a sensitivity approaching detection of one ma- lignant cell in 100,000 reactive lymphocytes. 2,3 The authors make no mention of attempted blockade of cellular F c receptors. This step is to eliminate the common error of spuriously detecting light chain restriction while in fact observing a phenome- non of passive cytophilic immunoglobulin absorbed onto nonlym- phoid B cells. In contrast, failed blockade of F c receptors on lymphoid cells may lead to false-negative results. Apparently no specific gating strategies were utilized. For in- creased sensitivity, the routine use of a B-cell gate (CD19 1 lineage gate) is recommended. In addition, application of other unique mark- ers known to be present on subsets of non-Hodgkin’s lymphoma (NHL) and chronic leukemia cells provides added resolution for the assessment of bone marrow involvement. Unique gating strategies should differ based on the type of NHL present at diagnosis (i.e., follicular lymphoma (CD191/CD101), small lymphocytic leukemia/ lymphoma and mantle cell lymphoma (CD191/CD51), etc). This 240 © 1999 American Cancer Society

Upload: ganesh-raj

Post on 06-Jun-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

CORRESPONDENCE

Bone Marrow Staging of Patients with Non-Hodgkin’s Lymphoma

Is Flow Cytometry a Useful Test?

Representing a busy flow cytometry laboratory with an annualworkload of approximately 1400 acute leukemia and malignant

lymphoma cases, we read the article by Naughton et al.1 with con-siderable interest. The authors concluded that “flow cytometry ofbone marrow aspirates has a limited role in the routine staging andfollow-up of patients with an established diagnosis of [non-Hodgkin’slymphoma] NHL.”1 However, this conclusion must be challengedbecause a number of methodologic flaws and inadequacies in theprocess, analysis, and interpretation are evident in this study.2,3

The authors used Ficoll-Paque (Pharmacia Biotech, Piscataway,NJ) processed specimens rather than recommended erythrocyte ly-sis.2 Ficoll-Paque centrifugation of bone marrow aspirates is known toresult in the selective loss of some cell populations. This phenomenonis related to the fact that neoplastic lymphoid cells may have adifferent buoyant density from normal lymphocytes and may not befound where anticipated in the gradient.2 The authors, possibly un-knowingly, thus may have reduced the number of potentially analyz-able malignant lymphoid cells in their samples, which may explain aportion of their staged cases with a negative flow result.

Single color flow cytometry was used on 81 specimens and twocolor flow cytometry on the remaining 192 specimens. It currently iswidely accepted that even two color flow cytometric analysis is inad-equate for an appropriate characterization of hematologic neoplasiaand virtually precludes an accurate assessment of minimal residualdisease. Five parameter analysis including forward and side scattersaccompanied by at least three color analysis should constitute theminimal standard for immunophenotypic analysis of lymphoprolif-erative diseases, with a sensitivity approaching detection of one ma-lignant cell in 100,000 reactive lymphocytes.2,3

The authors make no mention of attempted blockade of cellularFc receptors. This step is to eliminate the common error of spuriouslydetecting light chain restriction while in fact observing a phenome-non of passive cytophilic immunoglobulin absorbed onto nonlym-phoid B cells. In contrast, failed blockade of Fc receptors on lymphoidcells may lead to false-negative results.

Apparently no specific gating strategies were utilized. For in-creased sensitivity, the routine use of a B-cell gate (CD19 1 lineagegate) is recommended. In addition, application of other unique mark-ers known to be present on subsets of non-Hodgkin’s lymphoma(NHL) and chronic leukemia cells provides added resolution for theassessment of bone marrow involvement. Unique gating strategiesshould differ based on the type of NHL present at diagnosis (i.e.,follicular lymphoma (CD191/CD101), small lymphocytic leukemia/lymphoma and mantle cell lymphoma (CD191/CD51), etc). This

240

© 1999 American Cancer Society

allows not only for immunophenotypic demonstrationof light chain restriction, but also assists in furthersubclassifying the type of NHL/chronic leukemia in-volving the bone marrow.

Differential diagnosis of bone marrow hemato-gones from surface immunoglobulin negative malig-nant B cells was not mentioned in the article byNaughton et al. The article states that “flow cytometricresults were considered positive if $50% gated B-cellslacked surface immunoglobin, CD19, or CD20.”1 Bonemarrow hematogones are immature lymphoid cells ofB-cell origin that characteristically lack surface immu-noglobulins but are CD191 and may coexpress CD20.Hematogones are not malignant, but their numbermay increase markedly in reactive, postinfectious,postchemotherapy restaged bone marrow. It is virtu-ally impossible to differentiate these hematogonesfrom neoplastic and surface immunoglobulin negativecells of follicular center cell lymphoma without fourcolor CD19/CD10/k/l analysis.

Representing a Canadian healthcare provider, wealso cannot help but comment on the cost of the flowcytometric analysis in the published article. The au-thors quote the amount of 1720 1995 $US per case fora panel of 14 antibodies. For a 25% of this quotedamount our laboratory performs more sophisticatedchronic leukemia/NHL workup panels that are com-prised of twice as many antibodies. Our costs, whichamount to 450 $Cdn, include reagent (33 antibodypanel), and technical and professional fees. Tradi-tional staging of a bone marrow aspirate and biopsy inour institution is estimated to cost 300 $Cdn (includ-ing nursing, technical, and professional fees). Theabove cost analysis makes flow cytometry a very sen-sitive and yet cost-effective method of staging NHL inour center.

In contrast to the published article, our laboratoryhas yet to see a single case that was considered posi-tive by a standard morphologic assessment of bonemarrow aspirate and biopsy and yet negative on flowcytometry. However, we do encounter cases routinelythat are designated as either morphologically nega-tive, atypical, or showing the presence of “nonspecificlymphoid aggregates” that, without immunopheno-typic analysis, would spuriously have been considerednegative, missing the actual minimal residual disease.If appropriately performed and interpreted, flow cyto-metric analysis may indeed replace the traditionalmethod of staging NHL patients.

REFERENCES1. Naughton MJ, Hess JL, Zutter MM, Bartlett NL. Bone mar-

row staging in patients with non-Hodgkin’s lymphoma. Isflow cytometry a useful test? Cancer 1998;82:1154 –9.

2. Stelzer GT, Marti G, Hurley A, McCoy P Jr., Lovett EJ,Schwartz A. U.S.-Canadian Consensus recommendations onthe immunophenotypic analysis of hematologic neoplasiaby flow cytometry: standardization and validation of labo-ratory procedures. Cytometry 1997;30:214 –30.

3. Borowitz MJ, Bray R, Gascoyne R, Melnick S, Parker JW,Picker L, et al. U.S.-Canadian Consensus recommendationson the immunophenotypic analysis of hematologic neopla-sia by flow cytometry: data analysis and interpretation. Cy-tometry 1997;30:236 – 44.

Iwona Auer, M.D.Joanne Luider, B.Sc.

Flow Cytometry LaboratoryCalgary Laboratory Services

Foothills Medical Centre,Calgary, Alberta, Canada

Author Reply

W e appreciate the comments and recommenda-tions of Dr. Auer and Ms. Luider. Although

Ficoll-Paque centrifugation may lead to occasionalloss of neoplastic lymphoid cells, erythrocyte lysisalso may result in the loss of neoplastic populations.Fc receptor binding is less of a problem with Ficoll-Paque-purified cells. However we also use serum toblock Fc receptors. The second, fourth, and fifthparagraphs all speak to potential advantages ofthree or four color flow cytometry over two colorflow cytometry. We agree that five parameter anal-ysis accompanied by three color analysis appears tobe more sensitive. However, this method was notthe standard of care at the time these specimenswere analyzed, and still has not been adopted uni-versally. There were no consensus recommenda-tions before 1997. We believe our published resultsof the utility of two color flow cytometry for evalu-ating bone marrow aspirates in non-Hodgkin’s lym-phoma (NHL) specimens provides useful informa-tion for clinicians. Because the incidence of surfaceimmunoglobulin negative low grade lymphomas isvery low, difficulties in differentiating neoplasticand nonneoplastic cells are rare.

Although the comparison of charges for flowcytometric analysis at a U. S. versus a Canadianinstitution is interesting, the “true cost” of patho-logic and immunophenotypic staging is difficult toassess, particularly when comparing health care sys-tems financed through completely different mecha-nisms. In the U. S., reimbursement through third-party payers generally is only a fraction of the

Correspondence 241

charges. The charges were included only to pointout that this is a relatively expensive test, which inour hands added nothing to routine morphologicexamination.

Dr. Auer and Ms. Luider’s suggestion that “flowcytometric analysis may indeed replace the tradi-tional method of staging NHL patients” is not wellsupported by our study, nor by the results of Coad etal.,1 who found highly sensitive polymerase chainreaction (PCR) was negative in 43% of pathologicallypositive cases. As with PCR, the sensitivity of flowcytometry, regardless of the number of colors, willbe limited by the failure to aspirate sufficient lym-phoma cells. This particularly is true of follicularcenter cells, which often are embedded in reticulinfibers adjacent to bone trabeculae. To our knowl-edge, there are no published reports of the 100%sensitivity of flow cytometry for bone marrow stag-ing in NHL, which Dr. Auer and Ms. Luider quotefrom their laboratory. This is an important issue andwe would be interested in the results of a studysimilar to ours applying three color flow cytometryto bone marrow staging for NHL. Dramatically moresensitive results would provide an impetus foradopting the U. S./Canadian recommendations.

REFERENCES1. Coad JE, Olson DJ, Christensen DR, Lander TA, Chibbar R,

McGlennen RC, et al. Correlation of PCR-detected clonalgene rearrangements with bone marrow morphology in pa-tients with B-lineage lymphoma. Am J Surg Pathol 1997;21:1047–56.

Michael J. Naughton, M.D.Division of Medical Oncology

Department of MedicineWashington University School of Medicine

St. Louis, MissouriJay L. Hess, M.D., Ph.D.

Mary M. Zutter, M.D.Lauren V. Ackerman Laboratory

of Surgical PathologyDepartment of Pathology

Washington University School of MedicineSt. Louis, Missouri

Nancy L. Bartlett, M.D.Division of Medical Oncology

Department of MedicineWashington University School of Medicine

St. Louis, Missouri

Long Term Follow-Up andLate Complications of2-Chlorodeoxyadenosine inPreviously Treated, Advanced,Indolent Non-Hodgkin’s Lymphoma

We read with interest the article by Kong et al.1

reporting the long term follow-up and late com-plications of 2-chlorodeoxyadenosine (2-CdA) for pa-tients with previously treated, advanced, indolentnon-Hodgkin’s lymphoma. Herein we report the re-sults of 2-CdA treatment for patients with chroniclymphoproliferative disorders seen at our institution.

Thirty-six patients, 28 males and 8 females, wereenrolled in the study between May 1993 and May 1998at the Department of Hematology-Oncology, Univer-sity of Ankara Medical School, Ankara, Turkey. Thepatients’ median age was 56.3 years. Fourteen of themhad chronic lymphocytic leukemia (CLL), 15 had lowgrade non-Hodgkin’s lymphoma (NHL), and 7 hadhairy cell leukemia (HCL). 2-CdA therapy was primarytherapy for 6 patients (5 HCL and 1 NHL), and theremaining patients who were unresponsive to previ-ous therapy received it as second-line therapy. 2-CdAwas given as a continuous intravenous infusion at adose of 0.1 mg/kg/day for 7 days until achievement ofcomplete response (CR). Patients received 1– 4 coursesof 2-CdA therapy (the median number of courses was2). Thirty-three patients were given prophylactic anti-viral therapy (acyclovir, 200 mg 3 5/day) for a mini-mum of 4 weeks.

At the time of the current analysis, 14 of 36 pa-tients (38.9%; 5 HCL, 3 CLL, and 6 NHL) had achievedCR, 14 of 36 (38.9%; 2 HCL, 5 CLL, and 7 NHL) hadpartial response (PR), and 8 of 36 (22.2%; 6 CLL and 2NHL) had a stable outcome. Two NHL patients withCR, 1 NHL patient with PR, and 1 patient with stableCLL experienced relapse between 1 and 10 months.Treatment was generally well tolerated. Grade 4 my-elosuppression was seen in 19.4% of patients (3 CLL, 2NHL, and 2 HCL), of whom 3 developed febrile neu-tropenia. Myelodysplasia was reported by Orchard etal.,2 but it was not observed in any of our cases.Exacerbation of herpes zoster infection was observedin three patients who did not receive antiviral prophy-lactic therapy. Currently, after a follow-up period of6 –38 months (median follow-up, 13.2 months), 31 ofthe patients are still alive, and the median disease freesurvival is 17.5 months for HCL, 21 months for CLL,and 17.3 months for NHL. Because HCL was the mostcommon diagnosis in the treatment group, the overall

242 CANCER January 1, 1999 / Volume 85 / Number 1

response rate was similar with first-line treatment(83.3%, 5 of 6 patients, vs. 76.7%, 23 of 30 patients).Five patients who were given second-line therapy dieddue to non-therapy-related adverse effects, i.e., myo-cardial infarction from preexisting coronary heart dis-ease (n 5 2) and pulmonary embolism (n 5 1).

2-CdA is an effective agent for primary and/orsecondary treatment of patients with previouslytreated or advanced chronic lymphoproliferative dis-orders.3– 8 Based on the results of our current study, wealso believe that 2-CdA is both effective and nontoxic,suggesting its possible use alone and also as first-linetherapy for patients with HCL, NHL, or CLL.

REFERENCES1. Kong LR, Huang CF, Hakimian D, Variakojis D, Klein L,

Kuzel TM, et al. Long term follow-up and late complicationsof 2-chlorodeoxyadenosine in previously treated, advanced,indolent non-Hodgkin’s lymphoma. Cancer 1998;82:957– 64.

2. Orchard JA, Bolam S, Oscier DG. Association of myelodys-plastic changes with purine analogues. Br J Haematol 1998;100:677–9.

3. Rondelli D, Lauria F, Zinzani PL, Raspadori D, Ventura MA,Galieni P, et al. 2-chlorodeoxyadenosine in the treatment ofrelapsed/refractory chronic lymphoproliferative disorders.Eur J Haematol 1997;58:46 –50.

4. Mercieca J, Matutes E, Emmett E, Coles H, Catovsky D.2-chlorodeoxyadenosine in the treatment of hairy cell leu-kemia: differences in response in patients with and withoutabdominal lymphadenopathy. Br J Haematol 1996;93:409 –11.

5. Saven A, Piro LD. Complete remissions in hairy cell leuke-mia with 2-chlorodeoxyadenosine after failure with 29-de-oxycoformycin. Ann Intern Med 1993;119:278 – 83.

6. Brugiatelli M, Holowiecka B, Dmoszynska A, Krieger O,Planinc-Peraica A, Labar B, et al. 2-chlorodeoxyadenosinetreatment in non-Hodgkin’s lymphoma and B-cell chroniclymphocytic leukemia resistant to conventional chemother-apy: results of a multicentric experience. Ann Hematol 1996;73:79 – 84.

7. Beksac M, Ilhan O, Uysal A, Koc H, Idilman R, Arat M, et al.Cladribine therapy in chronic lymphoproliferative disor-ders: single center experience. ISH-EHA Combined Haema-tology Congress, Amsterdam, the Netherlands, July 4 – 8,1998:244.

8. Piro LD. Purine nucleoside therapy of low-grade follicularlymphoma. Ann Oncol 1996;7:S41–S47.

Meral Beksac, M.D.Mutlu Arat

Department of Hematology-OncologyUniversity of Ankara Medical School

Ankara, TurkeyRamazan Idilman

Department of GastroenterologyUniversity of Ankara Medical School

Ankara, Turkey[The authors were invited to reply and declined theinvitation.]

Lymph Node Harvest Reporting inPatients with Carcinoma of theLarge Bowel

A French Population–Based Study

The article by Maurel et al.1 in a recent issue ofCancer raised several questions and points for dis-

cussion. The authors analyzed reports of specimensfrom 851 colorectal carcinoma patients and reachedthe conclusion that at least 8 lymph nodes should beexamined histologically to provide a reliable indica-tion of negative lymph node status.

There are several recommendations in the worldliterature regarding the number of lymph nodes thatshould be recovered for adequate staging of lymphnode negative cases. Some articles state that a mini-mum of 10 lymph nodes should be examined,2 butothers suggest as many as 123 or even 17 lymphnodes.4 Many of these conclusions are based on math-ematic models.

It is evident that at least one lymph node must beharvested to identify a lymph node metastasis, andthis is why I find it difficult to understand how 65patients in whom no lymph nodes were recoveredcould be classified as lymph node negative for metas-tasis.1 These cases belong in the unclassifiable (pNx)category.

It also appears obvious that the chance of identi-fying a lymph node metastasis increases with thenumber of lymph nodes examined, but this latterstatement may lose its validity after a certain numberof lymph nodes have been examined. The number oflymph node positive specimens in this study increasedsignificantly with the number of lymph nodes exam-ined.1 Not all, but many lymph node metastases areenlarged and thus are easier to recover from the peri-colic fat. This could explain in part why more lymphnodes were identified in cases with metastases.

It also is common in these mathematic modelsthat there always is a substantial number of cases thatfall into the “less than optimal” category (not enoughlymph nodes examined). Using mathematic models, ahigher average number of lymph nodes harvested in astudy population will result in a higher minimal num-ber of lymph nodes required to be recovered. We faceda similar problem with axillary lymph node metastasesfrom breast carcinoma. With an average of sevenlymph nodes per axillary lymph node specimen, wereached the conclusion that at least ten lymph nodesmust be examined for a reliable classification of a caseto the pN0 category. Greater than 50% of the cases had

Correspondence 243

a “less than optimal” number of lymph nodes exam-ined.5 We then audited the dissection process andraised the average number of lymph nodes requiredper specimen to 22. By applying the same mathematicmodel to these cases the suggested minimal numberof lymph nodes to be investigated increased to 27,with approximately 50% of the specimens becoming“less than optimal” on the basis of the new criteria,but all having 10 – 42 lymph nodes examined.

The problem of the lymph nodes being selectedrandomly also was raised in the article.1 However,only the lack of identification of the apical or pedicularlymph node was addressed. Only peritumoral lymphnodes have a higher chance to be involved by a me-tastasis. If the majority of the lymph nodes come fromwithin a certain distance from the tumor, increasingtheir number most likely would not influence the de-tection rate of metastatic disease, but would appear tomake the specimen acceptable concerning the num-ber of lymph nodes examined.

I completely agree with the authors’ last conclu-sion stating that the number of lymph nodes re-ported to be negative in pN0 patients should beused as a stratification criterion in clinical studies,but if this is the case, strict protocols stating wherethese lymph nodes should come from also could beincluded. Six lymph nodes might be sufficient, asshown by the survival study cited in the article.1

Eight,1 ten,2 or more3,4 lymph nodes may be better.Not only should surgeons provide pathologists withat least eight to ten or more lymph nodes for min-imizing the N0 misclassification, pathologists alsoshould make efforts to find this number. This at-tempt by pathologists may be a more significantfactor in lymph node recovery rates than the surgi-cal procedures themselves.

REFERENCES1. Maurel J, Launoy G, Grosclaude P, Gignoux M, Arveux P,

Mathieu-Daude H, et al. Lymph node harvest reporting inpatients with carcinoma of the large bowel. A French pop-ulation-based study. Cancer 1998;82:1482– 6.

2. Jass JR. Prognostic factors in rectal cancer. Eur J Cancer1995;31A:862–3.

3. Fielding LP, Arsenault PA, Chapuis PH, Dent O, Gathright B,Hardcastle JD, et al. Clinicopathological staging for colorec-tal cancer: an International Documentation System (IDS)and an International Comprehensive Anatomical Terminol-ogy (ICAT). J Gastroenterol Hepatol 1991;6:325– 44.

4. Goldstein NS, Sanford W, Coffey M, Layfield LJ. Lymph noderecovery from colorectal resection specimens removed foradenocarcinoma. Trends over time and a recommendationfor a minimum number of lymph nodes to be recovered.Am J Clin Pathol 1996;106:209 –16.

5. Cserni G. Changing trends in lymph node recovery fromaxillary clearance specimens in breast cancer: possible im-

plications for the quantitative axillary status from a 17 yearretrospective study. Eur J Oncol 1997;2:403– 8.

Garbor Cserni, M.D.Department of Pathology

Bacs-Kiskun CountyTeaching Hospital

Kecskemet, Hungary

Author Reply

We thank Dr. Cserni for his interest and commentson our article. We would like to develop certain

points he discussed. In France, where colorectal car-cinoma is managed with a decentralized health caresystem, studies of health care practices must be pop-ulation-based to be representative. In that way, ourstudy is free of the types of selection biases that usu-ally limit conclusions drawn from data collected inspecialized centers.

Contrary to Dr. Cserni’s suggestion, we did notapply a preexisting mathematic model to our data.First, we established a sample of resected colorectalcarcinomas representative of routine and currentpractices in France and then we investigated by logis-tic regression the relation between the regional lymphnode classification of the tumor (i.e., N1 vs. N2) andthe number of lymph nodes examined. Dr. Cserniwondered how 65 patients in whom no lymph nodeswere recovered could be classified; in fact, they wereexcluded from further analysis.

We agree with Dr. Cserni that not all but manylymph nodes with metastases are enlarged and thuseasier to recover from the pericolic fat. In spite of theinfluence of many factors such as the variability of thehost-tumor response, the heterogeneity of surgicaland pathologic practices is likely to be the chief expla-nation for the variations in the number of lymphnodes examined, with the mean value for each depart-ment showing a wide range in our study, as it did foreach hospital in the Large Bowel Cancer Projectstudy.1

With regard to the “less than optimal” category(not enough lymph nodes examined), at least twosurvival studies (Givel et al., unpublished data and ref.2) already have demonstrated an independent effectof the number of lymph nodes examined, with a pla-teau number very close to the one determined in ourstudy.

One limitation of our study was the lack of avail-able precision regarding the location of positive lymphnodes; however, after the completion of our work we

244 CANCER January 1, 1999 / Volume 85 / Number 1

observed that the latest edition (1997) of the TNMstaging system did not take into account the locationof positive and examined lymph nodes.

Finally, the validity of our statement that at leasteight lymph nodes should be examined to assesslymph node status correctly in colorectal carcinomapatients had been reinforced by the most recentFrench Consensus Meeting.3

REFERENCES1. Blenkinsopp WK, Stewart-Brown S, Blesovsky L, Kearney G,

Fielding LP. Histopathology reporting in large bowel cancer.J Clin Pathol 1981;34:509 –13.

2. Michiels C, Boutron MC, Chatelain N, Quipourt V, Roy P,Faivre J. Facteurs pronostiques des adenocarcinomes colo-rectaux de stade B de Dukes. Etude d’une serie de popula-tion. Gastroenterol Clin Biol 1994;18:466 –71.

3. Conference de Consensus. Conclusions et recommenda-tions du Jury. Gastroenterol Clin Biol 1998;22:S275– 88.

Jean Maurel, M.D., Ph.D.Service de Chirurgie Digestive-CHu

Caen et Registre des TumeursDigestives du Calvados

CJF INSERMCaen, France

Telomerase Activity inPrecancerous Hepatic Nodules

D iagnosis of hepatocellular carcinoma (HCC) ofsmall, differentiated type at an early stage is some-

times difficult even by imaging procedures, biochem-ical tumor markers, or histologic examinations. Wepreviously reported that the appearance of telomeraseactivity was closely correlated with the incidence ofHCC in hepatic tissues.1,2 The cutoff level of telomer-ase activity, by using semiquantitative assay, for itspractical usage in HCC diagnosis, was tentatively setfor 0.6 mg liver protein per assay at 10-cell equivalentactivity of a gastric carcinoma cell line, MKN-1.2 Thiscutoff level was twice the highest activity observed innontumor chronic liver diseases, and telomerase pos-itive incidence above the cutoff level was 85% in allHCCs and 68% in well-differentiated HCCs (Table 1).In well-differentiated HCCs, characteristic appear-ances are seldom revealed by imaging techniques, andtumor markers are usually normal or only minimallyelevated.3 Therefore, it is clear that telomerase activitymay be a useful tool for precise and early diagnosis ofsmall, well-differentiated HCC.

However, Hytiroglou et al. recently reported that1) there was a clear-cut difference in telomerase ac-

tivity level between HCC and cirrhotic liver samples,but 2) the majority of large, noncancerous hepaticnodules, such as dysplastic lesions and large regener-ative nodules, had telomerase activity levels similar toHCC; therefore, they concluded that telomerase activ-ity is not always indicative of malignant transforma-tion of hepatic tissue.4

We will discuss here the apparent discrepancy inpublished data between theirs and ours. The firstpoint is the method used for telomerase activity. Te-lomeric repeat amplification protocol (TRAP) is a sen-sitive and simple procedure for measuring telomeraseactivity in tissue extracts.5 However, the conventionalTRAP method has two major limitations.6 One is poorquantitativeness. The conventional TRAP method hasa very narrow linearity range, probably due to thepolymerase chain reaction (PCR) amplification proce-dure. Some HCC tissues have telomerase activity lev-els several hundred times, or even more, higher thanprecancerous lesions. If an assay condition was usedunder which weak telomerase activity in precancerouslesions could be detected, most of the telomerase ac-tivity in carcinoma was saturated and underestimated.To measure telomerase activity semiquantitatively, se-rially diluted samples have been used. This titration isa kind of “end-point assay” and is not fully quantita-tive. Tenfold diluted samples did not always give 10-fold reduction of the signal intensity of electrophoreticpattern. The other problem with the conventionalTRAP method is the presence of inhibitory factorsagainst telomerase assay in some tissue extracts (butnot in cultured cells). Because this factor inhibits thePCR amplification step, two different methods wereused to eliminate the inhibitory effect. One was theaddition of an internal telomerase assay standard

TABLE 1Telomerase Activity Levels in Various Liver Tissues

Various liver tissuesPositivity abovethe cutoff levela

bNormal liver 0% (0/7)bChronic liver disease 0% (0/61)cPrecancerous hepatic nodule 30% (3/10)bHepatocellular carcinoma (HCC)

Well-differentiated 68% (15/22)Moderately differentiated 89% (39/44)Poorly differentiated 100% (12/12)Total 85% (66/78)

a Ten-cell-equivalent activity of a gastric-carcinoma cell line, MKN-1, set for practical usage in HCC

diagnosis.2

b Defined according to our report.2

c Newly assayed data; precancerous hepatic nodules were defined according to criteria of the Interna-

tional Working Party.3

Correspondence 245

(ITAS) DNA at the PCR step.6 After PCR amplificationand electrophoresis followed by densitometry, the in-tensity of telomerase product bands were normalizedby the intensity of the ITAS band. The second waselimination of inhibitory factor prior to PCR. DNAproducts from telomerase reaction were purifiedthrough phenol/chloroform extraction, and the puri-fied DNA was applied on PCR. Hytiroglou et al. usedthe conventional TRAP method to assay telomeraseactivity in hepatic tissues and did not mention havingthe above problems with this method.

The second point we wish to discuss here is thedifference in the samples employed. Although theirlarge nodular lesions corresponded histologically tothe definition of noncancerous hepatic nodules givenby the International Working Party,3,4 it was not clearwhether the lesions were detected by imaging tech-niques. When patients had chronic liver disease asso-ciated with a small nodular lesion in the liver duringcareful follow-up by imaging procedures, it was diffi-cult to distinguish between well-differentiated HCCand other noncancerous hepatic nodules, even by im-aging procedures, biochemical tumor markers, or his-tologic examination. This issue is a clinical problem.Clinically, when patients have chronic liver diseaseassociated with a hepatic nodular lesion that is remi-niscent of well-differentiated HCC, impairment ofliver function is caused by underlying cirrhosis, pre-cluding hepatectomy and permitting only fine-needlebiopsy. Whether detection of telomerase activity byneedle biopsy specimens can facilitate the diagnosisof malignancy is a matter of clinical interest.

We assayed telomerase activity quantitatively byboth modified TRAP method using ITAS and a newlydeveloped TRAP/HPA procedure.7 The latter proce-dure was recently published and had several advan-tages, such as 1) high sensitivity (telomerase activity in0.1 tumor cell equivalent could be detected), 2) a widerange of linearity (1–104 tumor cell equivalent), 3)practically no effect of tissue inhibitor, and 4) no needfor electrophoresis, densitometry, and normalizationof the gel pattern.7 Small, precancerous hepatic nod-ules associated with chronic liver disease (see Table 1)were defined by the International Working Party3 asfollows: less than 20 mm in greatest dimension; noappearance of classical HCC by imaging procedures,such as hypervascularity or capsule formation; no re-markable elevation of tumor markers, such as a-feto-protein or PIVKA-II; tissue samples composed ofsmall, eosinophilic hepatocytes with slightly increasedcellularity and without glandular structure. Thus, pre-cancerous hepatic nodules assayed here may corre-spond to dysplastic nodules proposed by the Interna-tional Working Party.3 Seven of 10 precancerous

hepatic nodules had telomerase activity below thecutoff level, and 3 of 10 were above the cutoff level.The activity in these 3 samples correspond to 21–MKN-1 cell equivalent, which was only twice the cut-off level and much lower than that in well-differenti-ated (mean 5 87 cell equivalent) and moderately orpoorly differentiated (mean $200 cell equivalent)HCC. These precancerous hepatic nodules with lowtelomerase activity may include the microscopic, well-differentiated HCC foci or immortalized noncancer-ous cells with a high risk of malignant transformationeven without definite histologic criteria of malignancy.This should be clarified in the future by a carefulexamination of in situ detection of telomerase activity.Currently, we believe that telomerase activity may stillbe useful as a tool for the diagnosis of well-differenti-ated HCC when an appropriate assay procedure isapplied for telomerase activity, although telomeraseactivity may not be an “all-or-none” marker of malig-nancy.

REFERENCES1. Tahara H, Nakanishi T, Kitamoto M, Nakashio R, Shay JW,

Tahara E, et al. Telomerase activity in human liver tissues:comparison between chronic liver disease and hepatocellu-lar carcinomas. Cancer Res 1995;55:2734 – 6.

2. Nakashio R, Kitamoto M, Tahara H, Nakanishi T, Ide T,Kajiyama G. Significance of telomerase activity in the diag-nosis of small differentiated hepatocellular carcinoma. Int JCancer 1997;74:141–7.

3. International Working Party. Terminology of nodular hepa-tocellular lesions. Hepatology 1995;22:983–93.

4. Hytiroglou P, Kotoula V, Thung SN, Tsokos M, Fiol MI,Papadimitriou CS. Telomerase activity in precancerous he-patic nodules. Cancer 1998;82:1831– 8.

5. Kim NW, Piatyszek MA, Prowse KR, Harley CB, West MD, HoPLC, et al. Specific association of human telomerase activitywith immortal cells and cancer. Science 1994;266:2011–5.

6. Wright WE, Shay JW, Piatyszek MA. Modifications of a telo-meric repeat amplification protocol (TRAP) result in in-creased reliability, linearity and sensitivity. Nucleic Acids Res1995;23:3794 –5.

7. Hirose M, Abe-Hashimoto J, Ogura K, Tahara H, Ide T,Yoshimura T. A rapid, useful and quantitative method tomeasure telomerase activity by hybridization protection as-say connected with a telomeric repeat amplification proto-col. J Cancer Res Clin Oncol 1997;123:337– 44.

Mikiya Kitamoto, M.D.First Department of Internal Medicine

Hiroshima University School of MedicineHiroshima, Japan

Toshinori Ide, Ph.D.Department of Cellular and Molecular Biology

Hiroshima University School of MedicineHiroshima, Japan

246 CANCER January 1, 1999 / Volume 85 / Number 1

Author Reply

We agree with Drs. Kitamoto and Ide that the con-ventional telomeric repeat amplification proto-

col (TRAP) assay has two major limitations: 1) it is notquantitative, and 2) additional controls are needed forverification of the specificity and sensitivity of theresults. To rule out false-negative results, we added toour samples (as a control for the polymerase chainreaction [PCR]) a diluted sample of IMR-32, a telom-erase activity positive neuroblastoma cell line.1 In ouropinion, this procedure, which has also been used byothers,2 is an acceptable alternative to using internaltelomerase assay standards (ITAS). To rule out false-positive results, we used RNase A, as generally ac-cepted; further, to ensure that the PCR reaction wouldstart at the same point in time in all samples, weperformed a variant of «hot start» PCR by adding Taqpolymerase and the CX primer after heating the sam-ples to stop telomerase activity. These modificationsof the conventional TRAP assay are described anddiscussed in sufficient detail in our article. However,an extensive discussion of TRAP assay methodologieswas beyond the scope of our article.

It is quite probable that, because of its sensitivity,the TRAP assay would be useful in clinical diagnostics,if adequately modified. In this sense, the method pro-posed by Hirose et al.3 is indeed very promising, be-cause it avoids the use of radioactive compounds andelectrophoresis and allows a relative quantitation oftelomerase activity. The results of this quantitation3,4

are in agreement with those of previous studies, inwhich the semiquantitative TRAP assay was used andshowed higher telomerase activity in hepatocellularcarcinomas (HCCs) than in noncancerous livers.5–7

However, according to every cutoff level used in Hi-rose et al.3 and Nakashio et al.4 (10 cell equivalent ofthe MKN-1 cell line, or 0.42–1 U representing anequivalent of K562 cell activity), there was still someoverlap between the relative telomerase activity valuesin chronic liver disease and well- or moderately dif-ferentiated HCCs.

Drs. Kitamoto and Ide state that the relative te-lomerase activity level in the precancerous noduleswas much lower than that in the well-differentiatedHCCs they studied. They also make reference to al-ready published results of the same research group,4

mentioning a very high mean value (87 cell equiva-lent) for the well-differentiated HCCs. However, thisvalue does not appear in the articles cited and, to ourunderstanding, cannot be concluded from the presen-tation of the respective results. For example, in Figure2 of Nakashio et al.4 the relative telomerase activity

value was below a 65 MKN-1 cell equivalent in all butone of the well-differentiated HCCs, whereas in Figure7B of Hirose et al.3 it was around or below 1 U (10 K562cell equivalent) in several well- and even moderatelydifferentiated HCCs.

Most importantly, and in contrast to what is foundin the sample groups of noncancerous livers, the rangeof the relative telomerase activity levels (correspond-ing either to MKN-1 or K562 cell equivalents) in thewell- and moderately differentiated HCC groups isvery broad.3,4 This means that the standard deviationsin the HCC groups, especially in the group of thewell-differentiated tumors, will also be very broad, notallowing mean values to be used to describe telomer-ase activity data. Moreover, exactly because of thewide range of telomerase activity levels observed inwell-differentiated HCCs, which extend into those ofnoncancerous chronic liver disease, overlap of levelsbetween dysplastic nodules and well-differentiatedHCCs is justified in the lower area of the scale. Thus,we do not think that there is any major discrepancybetween our findings and those of Drs. Kitamoto andIde, even though identifying the same value in all 3positive precancerous nodules (21 cell equivalent foreach of them) is quite surprising.

In terms of the histologic material utilized in ourstudy, all noncancerous nodules were detected oncareful macroscopic examination of livers removed attransplantation. Therefore, these lesions may bethought of as incidental findings, although we cannotrule out the possibility that at least some of themwould be detectable by radiologic means. A represen-tative area of each nodule was snap-frozen for molec-ular studies, and the remaining tissue was submittedfor histologic examination.1

It should be emphasized that, despite difficulties,histologic examination currently represents the goldstandard for diagnosis and classification of nodularhepatic lesions.8 Examination of fine-needle aspirates,even if combined with radiologic and serologic infor-mation, as suggested by Drs. Kitamoto and Ide, shouldnot be regarded as a substitute for histologic exami-nation at this time. It should also be emphasized thata significant number of large regenerative and precan-cerous nodules should be evaluated by the method ofHirose et al.3 before any conclusions regarding itspractical utility can be reached. Once the telomeraseactivity range in the various types of nodules is estab-lished, evaluation of such activity in fine-needle aspi-rates may become a useful auxiliary diagnosticmethod.

In conclusion, there is no doubt that the rapidlydeveloping methodologies of telomerase activity de-termination have significant potential for diagnostic

Correspondence 247

applications. Nevertheless, until the usefulness ofthese procedures is validated in large clinical studies,enthusiasm should be moderated.

REFERENCES1. Hytiroglou P, Kotoula V, Thung SN, Tsokos M, Fiel MI,

Papadimitriou CS. Telomerase activity in precancerous he-patic nodules. Cancer 1998;82:1831– 8.

2. Wan M, Li WZ, Duggan BD, Felix JC, Zao Y, Dubeau L.Telomerase activity in benign and malignant epithelial ovar-ian tumors. J Natl Cancer Inst 1997;89:437– 41.

3. Hirose M, Abe-Hashimoto J, Ogura K, Tahara H, Ide T,Yoshimura T. A rapid, useful and quantitative method tomeasure telomerase activity by hybridization protection as-say connected with a telomeric repeat amplification proto-col. J Cancer Res Clin Oncol 1997;123:337– 44.

4. Nakashio R, Kitamoto M, Tahara H, Nakanishi T, Ide T,Kajiyama G. Significance of telomerase activity in the diag-nosis of small differentiated hepatocellular carcinoma. Int JCancer 1997;74:141–7.

5. Tahara H, Nakanishi T, Kitamoto M, Nakashio R, Shay JW,Tahara E, et al. Telomerase activity in human liver tissues:comparison between chronic liver disease and hepatocellu-lar carcinomas. Cancer Res 1995;55:2734 – 6.

6. Nousso K, Urabe Y, Higashi T, Nakatsukasa H, Hino N,Ashida K, et al. Telomerase as a tool for the differentialdiagnosis of human hepatocellular carcinoma. Cancer 1996;78:232– 6.

7. Miura N, Horikawa I, Nishimoto A, Ohmura H, Ito H, Hiro-hashi S, et al. Progressive telomere shortening and telomer-ase reactivation during hepatocellular carcinogenesis. Can-cer Genet Cytogenet 1997;93:56 – 62.

8. International Working Party. Terminology of nodular hepa-tocellular lesions. Hepatology 1995;22:983–93.

Vassiliki Kotoula, M.D.1,2

Prodromos Hytiroglou, M.D.1,3

Swan N. Thung, M.D.3

Maria Tsokos, M.D.2

M. Isabel Fiel, M.D.3

Constantine S. Papadimitriou, M.D.1

1 Department of PathologyAristotle University Medical School

Thessaloniki, Greece2 Laboratory of Pathology

Section for Pediatric Pathology and ElectronMicroscopy

National Cancer InstituteNational Institutes of Health

Bethesda, Maryland3 Lillian and Henry M. Stratton–Hans Popper

Department of PathologyMount Sinai School of Medicine

New York, New York

Utilization of Polymerase ChainReaction Technology in theDetection of Solid Tumors

We read with interest the review article by Raj etal.1 on the utilization of polymerase chain reac-

tion (PCR) technology in the detection of solid tumors.However, there appear to be several issues of concern.

The authors comment on the difficulty of mak-ing comparisons between studies due to the lack ofuniformity in protocol. In reporting the variousstudies, careful interpretation must be made of theresults, as the authors have calculated positivity persample rather than per patient, making comparisonbetween studies difficult. In our pilot study, towhich the authors refer,2 the rate of K-20 reversetranscription (RT)–PCR positivity per patient was infact 20% (3 of 15 patients).

In discussing genetic alterations, the authorsclaim that the t(11;22) translocation is not unique to aspecific tumor, citing the presence of such mutationsin Ewing’s sarcoma, peripheral neuroepithelioma,Askin’s tumor, and esthesioneuroblastoma. However,these four tumors are regarded as one and the same incurrent surgical pathologic texts3 and thus are ex-pected to harbor the same mutation.

The authors also appear to regard cytokeratinsand keratins as different entities, when in fact they arethe same. Furthermore, expression of all cytokeratinsis tissue specific, not tumor specific, as claimed forcytokeratin 19 (K-19).

In contrast to the author’s statements, the studyby Burchill et al.4 did not look for cytokeratin 20 (K-20)expression in circulating melanoma cells, but in colo-rectal carcinoma.

With regard to breast carcinoma, there are nocurrently reported tissue specific or cancer specificmarkers that do not show expression in normal epi-thelial tissue, as far as we are aware. We and othershave previously reported the presence of cytokeratin19 (K-19) transcripts in normal lymph nodes,5 and wehave also recently detected the presence of maspingene expression in control lymph nodes, in contrast toLuppi et al.6

Establishing the association between surrogate tu-mor markers, such as cytokeratin expression in lym-phoid tissue and the actual presence of metastaticdisease, is vitally important. We have performed mu-tant allele specific amplification PCR (MASA) for K-rasin the lymph nodes of those patients who harbor K-rasmutations in their primary tumors. The results showgood concordance between cytokeratin 20 (K-20) ex-

248 CANCER January 1, 1999 / Volume 85 / Number 1

pression and mutant K-ras expression, confirmingthat K-20 RT-PCR does indeed detect occult colorectalcarcinoma cells (unpublished data).

Although there is strong evidence that PCR-based techniques enable very sensitive detection ofcirculating tumor cells and submicroscopic metas-tases, well-designed clinical studies are still re-quired to determine the clinical utility of these newdiagnostic approaches.

REFERENCES1. Raj GV, Moreno JG, Gomella LG. Utilization of polymerase

chain reaction technology in the detection of solid tumours.Cancer 1998;82:1419 – 42.

2. Gunn J, McCall JL, Yun K, Wright PA. Detection of micro-metastases in colorectal cancer patients by K19 and K20reverse-transcription polymerase chain reaction. Lab Invest1996;75:611– 6.

3. Rosai J. Soft tissue. In: Rosai J. Ackerman’s surgical pathol-ogy. 8th edition. Volume 2. St. Louis: Mosby, 1996:2105– 6.

4. Burchill SA, Bradbury MF, Pittman K, Southgate J, Smith B,Selby P. Detection of epithelial cancer cells in peripheralblood by reverse transcriptase–polymerase chain reaction.Br J Cancer 1995;71:278 – 81.

5. Yun K, Gunn J, Merrie A, Phillips LV, McCall JL. Keratin 19 isdetectable by RT-PCR in lymph nodes of patients withbreast cancer. Br J Cancer 1997;76:1112–3.

6. Luppi M, Morselli M, Bandieri E, Federico M, Marasca R,Barozzi P, et al. Sensitive detection of circulating breastcancer cells by reverse transcriptase–polymerase chain re-action of maspin gene. Ann Oncol 1996;7:619 –24.

A. E. H. MerrieDepartment of Surgery

Dunedin School of MedicineUniversity of Otago

Dunedin, New ZealandK. Yun

Department of PathologyDunedin School of Medicine

University of OtagoDunedin, New Zealand

J. L. McCallDepartment of Surgery

Faculty of Medicine and Health ScienceUniversity of Auckland

Auckland, New Zealand

Author Reply

We welcome the interesting points raised by Merrieet al. concerning our article on the utilization of

polymerase chain reaction (PCR) technology in thedetection of solid tumors, published this year in Can-cer.1

Merrie et al. point out correctly that we have re-ported the positivity per sample rather than per pa-tient. As discussed, this can be attributed to the lack ofuniformity of the reported data. While all reportsclearly stated the reverse transcriptase (RT)–(PCR)positivity per sample, only some also reported on theconcurrent positivity per patient. Clearly, the variabil-ity of the assay from patient to patient as well as forthe same patient at different periods of time is of greatinterest, as it would more accurately reflect the sensi-tivity and the reproducibility of the assay.

With regard to the t(11;22) translocations, Merrie etal. again correctly indicate that these tumors are now tobe regarded as a member of the family of small roundcell neoplasms of bone and soft tissue generically knownas primitive neuroectodermal tumors (PNETs). How-ever, some authors still prefer the distinction betweenthe various members of the PNET family.2 BecausePNET tumors exhibit solidarity in harboring the t(11;22)mutation, they would have a positive RT-PCR assay. Thisdenotes perhaps an absolute requirement for the t(11;22) mutation for tumorigenesis in these tumors. Of note,in an recent article, the t(11;22) mutation was reported tohave been detected in 2 of 12 typical neuroblastomas bythe PCR assay.3 Thus, we stand by our earlier contentionthat the t(11;22) mutation, while highly sensitive, is notrestricted to PNET tumors, and thus not absolutely spe-cific.

We agree that keratins and cytokeratins are thesame entity as classified by Moll’s catalogue system.We reported them as keratins or cytokeratins based onthe original author’s description.

The utility of various breast tumor markers is un-der active investigation and still evolving. Clearly, tra-ditional breast tissue markers lack specificity for thedisease. We are still not aware of any published datathat demonstrate maspin gene expression in normalperipheral blood.

Merrie et al. also report their unpublished data onthe detection of occult tumors. In our article, we pre-sented a state-of-the-art review of the utility of PCR inthe detection of solid tumors. Indeed, in the year sincethe acceptance of our article, more than 78 additionaloriginal articles have been published on the applica-tion of this technology for various tumors. This is anexciting area of research, and we strongly feel thatthese and ongoing studies will clarify the utility of PCRfor detecting solid tumors. We appreciate their inter-est in our research in this dynamic field.

REFERENCES1. Raj G, Moreno J, Gomella LG. Utilization of polymerase

chain reaction technology in the detection of solid tumors.Cancer 1998;82:1419 – 42.

Correspondence 249

2. Rosai J. Soft tissue. In: Rosai J. Ackerman’s surgical pathol-ogy. 8th edition. Volume 2. St. Louis. Mosby, 1996:1962– 4.

3. Burchill SA, Wheeldon J, Cullinane C, Lewis IJ. EWS-FLl1fusion transcripts identified in patients with typical neuro-blastoma. Eur J Cancer 1997;33:239 – 43.

Ganesh Raj, M.D., Ph.D.Division of Urology

Duke University Medical CenterDurham, North Carolina

Jose G. Moreno, M.D.Leonard G. Gomella, M.D.

Department of UrologyKimmel Cancer Center

Thomas Jefferson UniversityPhiladelphia, Pennsylvania

The Use of Epithelial MembraneAntigen and Silver-StainedNucleolar Organizer RegionsTesting in the Differential Diagnosisof Mesothelioma from BenignReactive Mesothelioses

Wolanski et al.1 recently studied the usefulness of acombined analysis of silver-staining nucleolar

organizer region (AgNOR) parameters and epithelialmembrane antigen (EMA) immunoreactivity to distin-guish between malignant and benign mesothelial tis-sues. The authors suggested the following diagnosticdecision-making system: Lesions should be consid-ered malignant when the AgNOR area is larger than0.6677 mm2 and/or when there is positive immunore-activity for EMA. We would like to discuss whetherthese results can be successfully applied in the routinework of other laboratories.

It is well known that variations of the technicalconditions, such as temperature, incubation time, andso on, may affect AgNOR staining,2 and this explainswhy in another study of benign and malignant me-sothelial proliferations the cutoff value for the AgNORarea was nearly twice3 the value determined byWolanski et al.1 Our own investigations2,4 – 6 confirmthe usefulness of the AgNOR staining for diagnosticpurposes, but we suggest a different method of deci-sion-making: After standardization of the AgNORtechnique, various parameters (e.g., AgNOR number,type, shape, and so on), which represent differentphysiologic cell functions,7 should be measured si-multaneously in cases with well-established diag-

noses. Combining all these parameters in a multivar-iate analysis (e.g., linear discriminant analysis), we canselect these parameters that best differentiate betweenthe diagnostic categories. In a second step, a combi-nation of these parameters may facilitate differentialdiagnosis. The system creates its own cutoff valuesbased on the “learning set,” and thus does not dependon data from other laboratories. After training thesystem, it will suggest the diagnosis in new cases. Wehave shown that this method works well even withrather simple AgNOR parameters, such as the numberof precipitations or the AgNOR type, which can berapidly and easily obtained and do not require imageanalysis systems.4 – 6 Furthermore, a decision-makingsystem based on different morphometric parametersis more reliable, because the influence of the individ-ual components is reduced.

In summary, we think that the use of multivariateanalysis systems can facilitate and improve diagnosticdecision-making based on AgNOR parameters underroutine conditions.

REFERENCES1. Wolanski KD, Whitaker D, Shilkin KB, Henderson DW. The

use of epithelial membrane antigen and silver-stained nu-cleolar organizer regions testing in the differential diagnosisof mesothelioma from benign reactive mesothelioses. Can-cer 1998;82:583–90.

2. Metze K, Lorand-Metze I. Interpretation of the AgNOR pat-tern in haematologic cytology. Acta Haematol 1993;89:110 –1.

3. Derenzini M, Nardi F, Farabegoli F, Ottinetti A, Roncaroli F,Bussolati G. Distribution of silver-stained interphase nucle-olar organizer regions as a parameter to distinguish neo-plastic from nonneoplastic reactive cells in human effu-sions. Acta Cytol 1989;33:491– 8.

4. Lorand-Metze I, Metze K. AgNOR clusters as a parameter ofcell kinetics in chronic lymphocytic leukemia. JCP Clin MolPathol 1996;49:357– 60.

5. Irazusta SP, Vassallo J, Magna LA, Metze K, Trevisan M. Thevalue of AgNOR and PCNA staining in endoscopic biopsiesof gastric mucosa. Pathol Res Pract 1998;194:33–9.

6. Cia EM, Trevisan M, Metze K. AgNOR technique: a helpfultool for the differential diagnosis in urinary cytology. Cyto-pathology In press.

7. Lorand-Metze I, Carvalho MA, Metze K. Relationship be-tween morphologic analysis of nucleolar organizer regionsand cell proliferation in acute leukemia. Cytometry 1998;32:51– 6.

Konradin Metze, Ph.D.Miriam A. Trevisan, Ph.D.

Department of PathologyFaculty of Medicine

State University of CampinasCampinas, Brazil

[The authors were invited to reply and declined theinvitation.]

250 CANCER January 1, 1999 / Volume 85 / Number 1

Stereotactic Radiosurgery for BrainMetastasis from Renal CellCarcinoma

The article of Mori et al.,1 who investigated stereo-tactic radiosurgery (rs) with or without whole-

brain radiotherapy (wbrt) for brain metastasis fromrenal cell carcinoma, merits comment.

The authors retrospectively evaluated the treat-ment results for 35 consecutive patients with 52 brainmetastases who underwent gamma knife rs. Twenty-eight patients (80%) additionally received wbrt, butonly 19 of them received wbrt with rs as a tumor boostfor first-line treatment. The others received wbrt atanother point in time, e.g., for progressive brain me-tastases. Thus, a rather inhomogenous patient group,divided into several small subgroups, was evaluated.However, because of the possible implications for fu-ture studies that would include patients with brainmetastases from other primary tumors as well, someinteresting results need to be reconsidered.

In general, overall median survival and local con-trol were comparable to previous analyses of rs thatalso included patients with nonrenal primary tu-mors.2,3 Although not yet demonstrated in a random-ized trial, surgical resection of brain metastases ap-pears to be of equal value.4 So far, both wbrt alone andwbrt plus non-rs boost do not show similarly favorableresults.5,6 Wbrt plus rs boost is now being tested ver-sus wbrt alone (Radiation Therapy Oncology Group[RTOG] 95-08).7 Thus, rs or surgical resection are con-sidered the current treatment of choice for patientswith favorable prognoses. These options should bediscussed with the patients to take their personal pref-erences into account. The role of adjuvant wbrt isbeing debated and prospectively evaluated in the on-going European Organization for Research and Treat-ment of Cancer (EORTC) 22952 study.7

Adjuvant wbrt (in conjunction with either rs orsurgical resection) has to be judged considering itsinfluence on overall survival, “distant” control (i.e.,any new brain metastases in a location distinct fromthe original lesion), local control (i.e., control in thesame site as the surgically or radiosurgically treatedoriginal lesion), toxicity/quality of life, and prolonga-tion of treatment time/cost. Overall survival is highlydependent on several prognostic factors, including ex-tracranial disease control.5,8 Furthermore, Mori et al.1

and my former group4 showed that salvage therapymight be effective for relapsing patients. Severalgroups observed that survival was not significantly

reduced by brain failure.4,9,10 Therefore, survival mightnot be a convincing rationale for adjuvant wbrt.

Mori et al.1 and Armstrong et al.11 did not find areduction of “distant” failures after wbrt. This failurerate might be influenced by extracranial disease activ-ity,4 i.e., new dissemination of tumor cells, rather thansurvival of clonogenic subclinical deposits present atthe time of wbrt. The latter hypothesis is consistentwith the high rate of patients with active visceral me-tastases (78%)1 and uncontrolled primary tumors (40 –50%).11 Therefore, reduction of distant failure mightbe achieved only in patients with controlled extracra-nial disease.

However, Mori et al.1 and Flickinger et al.3 showedthat wbrt increased local control compared with rsalone. This is a very interesting result that merits someradiobiologic comment. Table 1 shows calculations ofthe biologically effective dose (BED) according to thelinear-quadratic model,12 assuming values of a/b forrenal cell carcinoma published in the literature (4 –20gray [Gy]).13–15 According to Mori et al.,1 rs doses of 16Gy and 20 Gy (margin) were used for the calculations,because most of the patients received 16 Gy whenwbrt was applied and 20 Gy when rs alone was given.For wbrt, the common 30 Gy schedule was chosen.Assuming that treatment was applied without a rele-vant break between wbrt and rs, a simple calculationwithout correction terms was performed (BED 5 totaldose z (1 1 single dose/(a/b))). Thus, it can be shownthat a combination of wbrt and rs (30 plus 16 Gy) ismore effective than rs alone (20 Gy) without increas-ing the risk of late toxicity (a/b 5 2 Gy) at the site of rs.This increased effectiveness is more pronounced intumors with high a/b values. To provide equivalenteffectiveness, the margin dose of rs alone needed to beincreased, e.g., to 24 Gy (a/b 5 10 Gy), leading to ahigher risk of late toxicity. Compared with rs alone, acombined approach additionally utilizes the advan-tages of fractionated radiotherapy, e.g., reoxygenationand redistribution, and reduces the risk of marginalfailures due to incorrect delineation of the target vol-ume. However, on the one hand additional wbrt de-

TABLE 1Biologically Effective Dose (BED) Calculations According to theLinear–Quadratic Model

Schedule a/b 5 2 Gy a/b 5 4 Gy a/b 5 10 Gy a/b 5 20 Gy

1. 10 3 3 Gy 75 52.5 39 34.52. 1 3 16 Gy 144 80 42 293. 1 3 20 Gy 220 120 60 404. 1. 1 2. 219 132.5 81 63.55. Comparison of

4. with 3. 60% 110% 135% 159%

Correspondence 251

creases the risk of late toxicity at the site of rs, whereason the other hand its possible influence on neurocog-nitive and other brain functions is well known. Forexample, my former group reported an actuarial prob-ability of Grade I–III late toxicity (RTOG/EORTC) of42% after 2 years.4 With other wbrt schedules, such as15 3 2.5 Gy (applied in RTOG 95-08) or 20 3 2 Gy, BEDvalues for the normal brain are even higher. The ef-fectiveness of a less toxic schedule, such as 15 3 2 Gy,e.g., applied as prophylactic cranial irradiation forsmall cell lung carcinoma, seems questionable, espe-cially in the treatment of tumors that are radioresis-tant.

Thus, as long as a survival benefit with wbrt hasbeen demonstrated only in retrospective8 but not inrandomized trials, there are two arguments in favor ofit: an increase in the local control rate without anincrease in the local late toxicity rate, and possibly adecrease in the “distant” failure rate for patients with-out active extracranial disease. However, these have tobe weighed against toxicity and cost. We will probablynot be able to either introduce or decline routine wbrtwithout a thorough evaluation of the factor that ismost crucial for any patient suffering from an ad-vanced tumor with brain metastases: quality of life.We urgently need a prospective evaluation of thequestion of whether possible treatment failure (withsubsequent salvage therapy) after local therapy aloneor adverse effects of routine wbrt are more disadvan-tageous to the patients and/or the health care system.I suggest that future studies be stratified according toextracranial disease status.

REFERENCES1. Mori Y, Kondziolka D, Flickinger JC, Logan T, Lunsford LD.

Stereotactic radiosurgery for brain metastases from renalcell carcinoma. Cancer 1998;83:344 –53.

2. Alexander E III, Moriarty TM, Davis RB, Wen PY, Fine HA,Black PM, et al. Stereotactic radiosurgery for the definitive,noninvasive treatment of brain metastases. J Natl CancerInst 1995;87:34 – 40.

3. Flickinger JC, Kondziolka D, Lunsford LD, Coffey R, Good-man M, Shaw E, et al. A multi-institutional experience withstereotactic radiosurgery for solitary brain metastasis. Int JRadiat Oncol Biol Phys 1994;28:797– 802.

4. Nieder C, Schwerdtfeger K, Steudel WI, Schnabel K. Patternsof relapse and late toxicity after resection and whole-brainradiotherapy for solitary brain metastases. StrahlentherOnkol 1998;174:275– 8.

5. Nieder C, Berberich W, Nestle U, Niewald M, Walter K,Schnabel K. Relation between local result and total dose ofradiotherapy for brain metastases. Int J Radiat Oncol BiolPhys 1995;33:349 –55.

6. Murray KJ, Scott C, Greenberg HM, Emami B, Seider M, VoraNL, et al. A randomized phase III study of accelerated hy-perfractionation versus standard in patients with unre-sected brain metastases: a report of the Radiation Therapy

Oncology Group (RTOG) 9104. Int J Radiat Oncol Biol Phys1997;39:571– 4.

7. National Cancer Institute. Curr Clin Trials Oncol 1998;5:1–204.

8. Smalley SR, Laws ER, O’Fallon JR, Shaw EG, Schray MF.Resection of solitary brain metastasis. J Neurosurg 1992;77:531– 40.

9. Arbit E, Wronski M, Burt M, Galicich JH. The treatment ofpatients with recurrent brain metastases. Cancer 1995;76:765–73.

10. Bindal RK, Sawaya R, Laevens ME, Hess KR, Taylor SH.Reoperation for recurrent metastatic brain tumors. J Neuro-surg 1995;84:442– 8.

11. Armstrong JG, Wronski M, Galicich JH, Arbit E, Leibel SA,Burt M. Postoperative radiation for lung cancer metastaticto the brain. J Clin Oncol 1994;12:2340 – 4.

12. Joiner MC, Van der Kogel AJ. The linear-quadratic approachto fractionation and calculation of isoeffect relationships. In:Steel GG. Basic clinical radiobiology. London: Arnold, 1997:106 –22.

13. Syljuasen RG, Belldegrun A, Tso CL, Withers HR, McBrideWH. Sensitization of renal carcinoma to radiation usingalpha interferon (IFNA) gene transfection. Radiat Res 1997;148:443– 8.

14. Gioanni J, Zanghellini E, Mazeau C, Amiel J, Poustis-DelpontC, Lagrange JL, et al. CAL 54, une nouvelle lignee cellulairederivee d’un carcinome renal humain: caracterisation etradiosensibilite. Bull Cancer 1996;83:553– 8.

15. Leung SW, Mitchell JB, Al-Nabulsi I, Friedman N, NewsomeJ, Belldegrun A, et al. Effect of L-Buthionine sulfoximine onthe radiation response of human renal carcinoma cell lines.Cancer 1993;71:2276 – 85.

Carsten Nieder, M.D.Department of Experimental Radiation Oncology

The University of Texas M.D. Anderson Cancer CenterHouston, Texas

Author Reply

We thank Dr. Nieder for his comments regardingour report on the use of radiosurgery and whole-

brain radiation therapy for patients with brain metas-tases from renal cell carcinoma. Increasingly, the roleand benefit of whole-brain irradiation is being dis-cussed among clinicians. Because the value of whole-brain irradiation compared with surgical resection orto radiosurgery remained questionable, and because itresults in occasional late cognitive morbidity, its userequired further evaluation. We appreciate the effortmade by Dr. Nieder to compare radiosurgery doseswith fractionated radiation doses using mathematicmodels. We have done so in other experiments inwhich malignant gliomas were used.1 Dr. Nieder iscorrect in recommending evaluation of data from ran-domized trials before drawing conclusions regardingthe pros and cons of any treatment. We agree that the

252 CANCER January 1, 1999 / Volume 85 / Number 1

quality-of-life issue will be crucial to an evaluation ofthe roles of specific treatments. We now have severallong term survivors who have cognitive deficits relatedto either whole-brain radiation therapy or other treat-ments. Could these have been avoided? Our first goalis to prolong survival. Our second is to control localbrain disease, and our third is to improve long termquality of life. All of these goals are important, andstudies should be designed to evaluate all three pa-rameters.

REFERENCE1. Kondziolka D, Somaza S, Comey C, Lunsford LD, Classen D,

Pandalai S, et al. Radiosurgery and fractionated radiationtherapy: comparison of different techniques in an in vivo ratglioma model. J Neurosurg 1996;84:1033– 8.

Douglas Kondziolka, M.D.Department of Neurological Surgery

University of Pittsburgh Medical CenterPittsburgh, PennsylvaniaJohn C. Flickinger, M.D.

Department of Radiation OncologyUniversity of Pittsburgh Medical Center

Pittsburgh, Pennsylvania

Why Hereditary NonpolyposisColorectal Carcinoma PatientsAppear to Have Better Survival thanPatients with Sporadic ColorectalCarcinoma

We read with interest the recently published studyby Watson et al.1 and the accompanying editorial

by Dr. Bernard Levin.2 Hereditary nonpolyposis colo-rectal carcinomas (HNPCCs) account for approxi-mately 5– 6% of colonic tumors. They are inherited,autosomal dominant tumors with 80% penetrance.Hereditary nonpolyposis colorectal carcinomas arehistologically and clinically distinct from sporadic co-lonic adenocarcinomas.3–5 Watson et al. have substan-tiated the observations of other investigators thatHNPCC patients have better survival than in patientswith sporadic colorectal carcinoma.5

The biologic behavior of HNPCC also seems to bequite distinct from that of sporadic colon carcinomas.These tumors do not seem to follow the “adenoma-carcinoma sequence” of genetic events proposed byVogelstein. Whereas mutations of tumor suppressorgenes and protooncogenes (APC, p53, and K-ras) are

common in sporadic colorectal carcinomas, they aremuch less frequent in HNPCC.6,7 On the other hand,mutations of DNA mismatch repair genes are a com-mon feature of HNPCC.7 This high mutation rate isdetected as increased variability in DNA microsatel-lites, the so-called “microsatellite instability.” Tumorswith microsatellite instability are called replication er-ror (RER) positive. Most HNPCC cases are RER posi-tive and most sporadic cancers negative.

The lower stage and relative infrequency of me-tastasis at the time of diagnosis may be related to thebiologic behavior of these tumors. The Crohn’s disea-se–like lymphoid response suggests a host immuneresponse to these tumors, which may limit spread andinvasion, as suggested by Graham et al.8 Shibata hassuggested that the genomic instability may itself con-tribute to the less invasive nature of these tumors. Thehigh rate of mutations associated with microsatelliteinstability may interfere with essential functions of themalignant cells and limit their spread.9

We have recently reported immunohistochemicalstudies of severely dysplastic colonic polyps.10 Wefound that 14 of 15 polyps with invasive carcinomawere positive for the p53 protein compared with 13 of27 cases of severely dysplastic polyps without invasivecarcinoma (P 5 0.003). Other investigators have alsoreported an association between p53 mutation andmetastatic disease, and overall poor prognosis in colo-rectal carcinomas.11 Therefore, we believe the low fre-quency of p53 mutations in HNPCC may be an impor-tant biologic feature that determines their relativelybetter survival. We also wonder whether the microsat-ellite instability mitigates against p53 mutations.

REFERENCES1. Watson P, Lin KM, Rodriguez-Bigas MA, Smyrk T, Lemon S,

Shashidharan M, et al. Colorectal carcinoma survival amonghereditary nonpolyposis colorectal carcinoma family mem-bers. Cancer 1998;83:259 – 66.

2. Levin B. Why hereditary nonpolyposis colorectal carcinomapatients appear to have better survival than patients withsporadic colorectal carcinoma. Cancer 1998;83:201–2.

3. Mecklin JP, Jarvinen HJ. Clinical features of colorectal car-cinoma in cancer family syndrome. Dis Colon Rectum 1986;29:160 – 4.

4. Jass JR, Smyrk TC, Stewart SM, Lane MR, Lanspa SJ, LynchHT. Pathology of hereditary nonpolyposis colorectal cancer.Anticancer Res 1994;14:1631– 4.

5. Sankila R, Aaltonen LA, Jarvinen HJ, Mecklin JP. Better sur-vival rates in patients with MLHI-associated hereditary colo-rectal cancer. Gastroenterology 1997;110:682–7.

6. Losi L, Ponz de Leon M, Jiricny J, Di Gregorio C, Benatti P,Percesepe A, et al. K-ras and p53 mutations in hereditarynon-polyposis colorectal cancers. Int J Cancer 1997;74:94 – 6.

7. Yamamoto H, Sawai H, Weber TK, Rodriguez-Bigas MA,Perucho M. Somatic frameshift mutations in DNA mismatchrepair and proapoptosis genes in hereditary nonpolyposiscolorectal cancer. J Cancer Res 1997;58:997–1003.

Correspondence 253

8. Graham DM, Appleman HD. Crohn’s-like lymphoid reac-tion and colorectal carcinoma: a potential histological prog-nosticator. Mod Pathol 1990;3:332–5.

9. Shibata D, Peinado MA, Ionov Y, Malkhosyan S, Perucho M.Genomic instability in repeated sequences is an early so-matic event in colorectal tumorigenesis that persists aftertransformation. Nat Genet 1994;6:273– 81.

10. Sheikh R, Min B, Tesluk H, et al. Correlation of Ki-67, p53,Adnab-9 staining and ploidy with multiplicity, morphology,size and invasiveness in colorectal adenomas with severedysplasia [abstract]. Gastroenterology 1998;114:A 2061.

11. Pricolo VE, Finkelstein SD, Hansen K, Cole BF, Bland KI.Mutated p53 gene is an independent adverse predictor ofsurvival in colon carcinoma. Arch Surg 1997;132:371–5.

R. A. Sheikh, M.D.B. H. Min, M.D.R. Teplitz, M.D.H. Tesluk, M.D.

B. H. Ruebner, M.D.University of California–Davis Medical Center

Sacramento, CaliforniaM. J. Lawson, M.D.

University of California–Davis Medical Centerand Kaiser Permanente Medical Center

Sacramento, California[The authors were invited to reply and declined theinvitation.]

Pretreatment Surgical Staging ofPatients with Cervical Carcinoma

The Case for Lymph Node Debulking

I read this article with interest; its conclusions sug-gested that debulking macroscopically positive

lymph nodes might afford a survival benefit to pa-tients with bulky or locally advanced cervical carci-noma.

The authors suggest that the relatively good out-come for those with macroscopically resected lymphnodes in Group C was a result of lymph node debulk-ing and that these patients did as well as those withmicroscopically involved lymph nodes in Group B. Inexamining the numbers presented by the authors, itwould appear that the relatively good outlook forthose with macroscopically resected disease might bealternatively related to a disproportionate representa-tion of patients with other, better tumor-related prog-nostic factors.

The factors that need to be considered before aconclusion can be drawn regarding the benefit of re-section are as follows:

1. The proportion of patients with paraaortic versuspelvic lymph nodes in each group: In the groupwith microscopically involved lymph nodes thatwere resected, 67% had pelvic disease and 33%had paraaortic lymph nodes, whereas in the groupwith macroscopic disease that was resected, 73%had pelvic disease and only 27% had paraaorticlymph node disease.

2. The effect of stage of disease on those with “mi-croscopic resected” versus “macroscopic re-sected” lymph nodes. The patients in Group Cappeared to have an overrepresentation of Stage Iand IIA patients compared with Group B (49% vs.39%).

3. Was there a larger number of lymph nodes in-volved in Group B than in Group C?

It is also interesting to calculate the possible mag-nitude of benefit in resecting and radiating macro-scopically involved paraaortic lymph nodes. Of 258patients studied, 20 had macroscopic paraaorticlymph nodes and 9 appear to be alive, i.e., represent-ing a benefit of 3.5% (9 of 258) at a cost of seriousbowel or bladder complication in 10.5%, a further 2%requiring laparotomy for lymphocysts, and lymphed-ema in 18.4%.

Can the authors perhaps define a better therapeu-tic ratio for research and treatment of macroscopiclymph node disease by delineating the characteristicsof those cured, e.g., the stage of disease from whichthe survivors were drawn and the number of macro-scopic resected lymph nodes? This might decrease thenumber of unsuccessful interventions.

Gillian M. Thomas, B.Sc., M.D.Toronto-Sunnybrook Regional Cancer Centre

Division of Radiation OncologyToronto, Ontario, Canada

Author Reply

We appreciate the comments of Dr. Thomas re-garding our recent article.1 There are several

points we would like to make in response.

1. In regard to the proportion of patients in GroupsB and C with pelvic and paraaortic lymph node(LN) metastases, there is insufficient informa-tion in the text to allow complete extrapolationof these data. To clarify: In group B (patientswith microscopically positive LN), 95% (37 of 39)had positive pelvic LN only, 5% (2 of 39) had

254 CANCER January 1, 1999 / Volume 85 / Number 1

positive pelvic and paraaortic LN, and no pa-tients had positive paraaortic LN only. Con-versely, in Group C (patients with macroscopi-cally positive LN that were completelyresection), 59% (44 of 74) had positive pelvic LNonly, 38% (28 of 74) had both pelvic andparaaortic positive LN, and 3% (2 of 74) hadpositive paraaortic LN only. For completeness,in patients in Group D (unresectable LN), 30% (6of 20) had positive pelvic LN only, 70% (14 of 20)had both pelvic and paraaortic LN metastases,and no patients had paraaortic LN metastasesonly. The discrepancy between these numbersand those extrapolated by Dr. Thomas from thetext occurred because, although it is stated inthe text that there were 13 patients with micro-scopically positive paraaortic LN, 11 of thesepatients were in Group C and not Group B be-cause they had macroscopically positive but re-sected pelvic LN. The numbers presented abovetherefore represent the true proportions withinthe defined LN groups. Based on this informa-tion, we would have expected patients in GroupC to do worse than patients in Group B, but thiswas not the case.

2. As noted in the article, when the data were cor-rected for stage, “ . . . Group C demonstrated sig-nificantly better survival than Group B for bothlow (I and II) and high (III and IV) stage patients(P , 0.05).”1 In addition, there was no statisticaldifference in the number of low (IB and IIA) andhigh (IIB and higher) stage patients betweenGroups B and C (P 5 0.3).

3. The patients in Group C had a statistically signif-icant higher mean number of positive LN (6.0 vs.2.1, P 5 0.0006) than the patients in Group B.Patients in Group C also had a higher mediannumber of LN removed (3 vs. 1) than those inGroup B.

Dr. Thomas’s final point makes the assumptionthat the only patients who potentially benefit frompretreatment surgical staging are those with macro-scopically positive paraaortic LN. We believe, how-ever, that these are not the only patients who benefitfrom this intervention. In addition, patients with mac-roscopically positive pelvic LN metastases only shouldalso benefit from the debulking of LN too large to besterilized by standard radiation therapy.2 This point issupported by the findings in the original report fromthis institution by Downey et al., which also demon-strated improved survival for patients in Group C overthose in Group D (and similar survival for patients inGroups B and C) when the groups were defined by

pelvic LN status.3 Based on this, there were 73 patients(27%) who potentially benefited from the intervention.Of these, 34 (13%) are currently alive. Furthermore,the median survival for these 73 patients with macro-scopically resected pelvic and/or paraaortic LN was 54months.

We would also add to this group who potentiallybenefited those patients with microscopic paraaor-tic LN metastases, as a significant number of thesepatients would presumably have had a recurrence iftreated with pelvic radiation only (13 patients). Con-versely, for those who would propose extended fieldradiation for all patients, then the 133 patients withnegative LN would have potentially benefited fromthe intervention because they would be spared therisk of complications from extended field radiother-apy.

If we consider only patients with macroscopicallypositive paraaortic LN, the 5-year survival for patientswith macroscopically positive paraaortic LN treatedwith radiation only is as low as 17%.4 This yields asurvival advantage among our patients of 26% (43% vs.17% historical controls).

Based on these data, we continue to believe thatuntil a nonsurgical method is developed that canreliably and accurately identify patients with LNmetastases, all patients with cervical carcinomawho have no evidence of widespread or distant me-tastases should undergo pretreatment extraperito-neal “staging” in order to 1) identify those patientswith microscopic LN metastases who would benefitfrom extended field radiation; and 2) debulk mac-roscopically positive LN prior to definitive radio-therapy.

REFERENCES1. Cosin JA, Fowler JM, Chen MD, Paley PJ, Carson LF, Twiggs

LB. Pretreatment surgical staging of patients with cervicalcarcinoma: the case for lymph node debulking. Cancer 1998;82:2241– 8.

2. Wharton JT, Jones HW III, Day TG Jr., Rutledge FN, FletcherGH. Preirradiation celiotomy and extended field irradiationfor invasive carcinoma of the cervix. Obstet Gynecol 1977;49:333– 8.

3. Downey GO, Potish RA, Adcock LL, Prem KA, Twiggs LB.Pretreatment surgical staging in cervical carcinoma: thera-peutic efficacy of pelvic lymph node resection. Am J ObstetGynecol 1989;160:1055– 61.

4. Vigliotti AP, Wen BC, Hussey DH, Doornbos JF, Staples JJ,Jani SK, et al. Extended field irradiation for carcinoma of theuterine cervix with positive periaortic nodes. Int J RadiatOncol Biol Phys 1992;23:501–9.

Jonathan A. Cosin, M.D.Obstetrics, Gynecology and Women’s Health

University of MinnesotaMinneapolis, Minnesota

Correspondence 255