Endometrial neoplasia-Screening the high-risk patient JOSEPH H. CRAMER, M.D.C.M., F.R.C.S.(C.)* RAYMOND J. OSBORNE, M.D .. F.R.C.S.(C.) Toronto, Ontario, Canada Increased longevity and use of exogenous estrogens appear to have placed more women at risk of developing endometrial carcinoma. Too infrequently, a diagnosis is made in these women at the premalignant stage of their disease. We assessed cytologically, by means of the Milan-Markley helix technique, 170 women seen at our institutions. These patients were classified in one of the following groups: late menstruators, periclimacteric bleeders, women receiving hormone replacement therapy, and women with endometrial pathologic conditions at the time of curettage. The diagnostic accuracy of the helix technique was assessed. Histologic confirmation was obtained with either Vabra or formal curettage. Tolerance on the part of the patient was excellent. Discomfort sufficient to terminate the procedure occurred in two of 189 samples. There were no complications. Tissue adequate for diagnosis was obtained in 180 of 189 samples. Sensitivity was 97%, specificity was 96%, and predictive value of a positive result was 97%. No cases of adenocarcinoma failed to be detected. We conclude that the Milan-Markley helix technique is safe and dependable in assessing women at risk of developing uterine pathologic conditions. (AM. J. OBSTET. GYNECOL. 139:285, 1981.)
and its precursors exfoliate cells into the endometrial cavity. Various methods 1 ~ 6 of obtaining specimens for cytologic study have been described over the past 40 years. The problem of collecting and studying these cells was alluded to over 20 years ago. Nieburgs 7 wrote, "These difficulties in evaluation of endometrial cells are due to the fact that with most techniques of cell collection, the cells obtained have undergone post-exfoliative changes. When such cells have undergone post-exfoliative deterioration the differential diagnosis by microscopic evaluation may be difficult. Perhaps the development of techniques for the collection of abraded, rather than exfoliative cells, may produce more satisfactory results." The best method available at present for obtaining abraded cells and tissue is formal curettage. This diagnostic gvnecologic procedure is the one most freENDOMETRIAL CANCER
From the Ontario Cancer Institute, The Princess 1Hargaret Hospit<~l, and the Toronto General Hospital. Presented at the Thirtv-sixth Annual Meeting of the Society of Obstetricians and Gynaecologists of Canada, jasper, Alberta, Canada, june 10-i4, i980. Reprint requests: Dr. R.]. Osborne, 330-170 St. George Street, Toronto, Ontario, Canada M4R 2M8. *Gordon Richards Fellow. 0002-9378/811030285+04$00.40/0
© 1981 The C. V. Moshy Co.
quently performed, and it accounts for a large proportion of hospital bed usage and operating time. This procedure is not without inconvenience and risk. which is even more substantial when it is done under general anesthesia, and it is inappropriate in the repeated follow-up of high-risk patients. We have analyzed the usefulness, specificity, sensitivity, and predictive value of a simple outpatient method of endometrial sampling in women at high risk for endometrial abnormalities. The Milan-Markley (Mi-Mark) helix technique allows for the study of directly smeared tissue obtained from within the uterine cavity, thus eliminating filtering, centrifugation, and complex processing. The helix technique acquires living cells, and, at times, tiny surface microanatomic structures, both of which lend themselves effectively to cytologic study. The Mi-Mark helix procedure was performed on 170 women, and a total of 189 specimens was obtained. The results indicate that this method is a reliable alternative to other outpatient methods of endometrial sampling. Materia! and methods
One hundred and seventy women were selected from those attending a special endometrial cytology clinic at the Toronto General Hospital or from inpa-
285
Am
Table I. Distribution of patients by age Age(yr)
20-29 30-39 40-49 50-59 60-69 >70 Total
4
5
28 48 51
34 170
Table II. Distribution by risk group Risk group Postmenopausal bleeders Late menstruators Women on hormone replacement therapy Dysfunctional bleeders Patients with endometrial pathologic condition Total
No.
53 18
41 33 25 170
tients at The Princess Margaret Hospital, Toronto, Canada. They were sampled a total of 189 times by means of the Mi-Mark technique. Their stratification by age is shown in Table I. The women sampled were a1nong those \Vho are generally considered to be at high risk for endometrial abnormalities. 11 In general, they fall into one of the following broad clinical groupings: ( 1) postmenopausal bleeders, (2) women on hormone replacement therapy, (3) late menstruators, (
ITIITI
plastic dilator handle of the
"paddle." The helix was twisted 10 times clockwise and was withdrawn as the twisting motion continued. The extruder-paddle was then used to remove the specimen and place it on a slide which was thoroughlv wetted
J
February l, ! 9K l Ohstet. Cmecol.
with 95'/c ethyl alcohol. Once on the slide, the specimen was spread evenlv. placed in a jar of Carnm·'s o;olution I(Jr ;~o minutes. fixed, and stained. The cytologic specimens obtained were all processed and examined bv a single pathologist at the Toronto General Hospital. Standard criteria for cvtologic atvpia were used in classifying these samples. Abnormal processes \Vere indicated by sorne ofrhe follo\\-ing feJiures: enlarged nuclei, multiple nucleoli, chromatin clumping, irregular nuclear borders, irregular thickening or budding nuclear membranes. nuclear crowding. altered nuclear-cytoplasmic ratios, ~md cytoplasmic \·acuoles. The degree of cellular abnormalitv was classified into benign hyperplasia, adenomatou: hyperplasia, and atypical adenomatous hyperplasia. \\ith the latter two being graded into mild. moderate, and scnTe abnormalities. Of these 170 women, 14H were sampled concurrently or within a month's time of helix sampling by either Vabra suction curettage or formal fractional curettage. Hysterectomy after curettage was carried out in nine patients, and pathologic studv was done ~tt the same institution. Direct comparison between the results of helix sampling and these other methods was then rnade.
Results
One hundred and seventy women were sampled 189 times by means of the helix. Pathologic confirmation bv means of one of the other described met hods was made in 148 instances. Patient compliance. Severe pain necessitated termination of two attempts at sampling. Local anesthetic was needed in eight of the 189 sampling procedures. Dilators were needed in seven of 189 procedures. 1\'o uterine perforations were recognized ro have occurred. ;\Jo patient experienced a syncopal episode, but 14 patients experienced nausea. ~o patient \·tnnitcd during or after the procedure. Slight bleeding from the uterus occurred in 49 out of 189 samplings. Tissue adequate for diagnosis was obtained in lt\2 of IH9 samplings. Of the 148 patients in whom a pathologic diagnosis was obtained, the following cvtologic diagnoses were made. Seventy-six helix samples were interpreted as showing no abnormal cells, :n had cells or cellular structures consistent with endometrial hvperplasia, 11 were consistent with mild atypical adenomatous endometrial hyperplasia. five with moderate. and fi \ e \vith severe. 'I\v·o santples \\·ere interpreted as showing benign squamous metaplasia, and nine cases were interpreted as adenocarcinoma of the endometrium. The diagnoses made on samples obtained ll\ curet-
Volume 139
Endometrial neoplasia-Screening high·risk patient 287
Number :l
Table III. Predictive value of the helix samples applied to healthy and diseased populations Helix No. with positive result No. with disease No. without disease Totals
or
76 TP
T:
X
No. with negative result
TP (69) FP (3) FP + TP (72)
Predictive value of a positive result: TP
74
I
FP x 100 or
FN (2) TN (74) TN+ FN (76)
~~
Totals TP + FN (71) TN+ FP (77) TP + FP + TN + FN ( 148)
x 100 = 96%. Predictive value of a negative result: FN :~N x 100
100 = 97.3%. true positive; FP = false positive; TN
= true negative;
tage are as follows: 77 patients had normal endometrium, 38 had endometrial hyperplasia, 13 had mild, five had moderate, and six had severe atypical adenomatous hyperplasia; nine cases were reported to be adenocarcinoma of the endometrium. Two cases of abnormal endometrium failed to be detected by the helix: one was interpreted as mild atypical adenomatous hyperplasia, and the other as severe atypical adenomatous hyperplasia. In these cases, the helix cytologic findings were reported as, "no abnormal cells present." Three cases were found to be abnormal by the helix sampling method. In all three, the diagnosis was "groups of cells consistent with focal, mild endometrial hyperplasia." Curettage specimens were reported to be normal in these three cases. All nine cases of adenocarcinoma diagnosed pathologically were correctly diagnosed by the helix cytologic specimen.
Comment The Mi-Mark helix technique was developed in 1973, for sampling the endometrial cavities of women on an outpatient basis, with little discomfort, low cost, and minimal risk. As our results indicate, these practical considerations have been met. Only two attempts at sampling could not be carried out. No serious sequelae or side effects occurred. Compliance and tolerance on the part of the patients were excellent. Tissue adequate for diagnosis was obtained in 182 of 1159 samplings, or 96.3%. This represents results far superior to those recently quoted in the literature, 9 and certainly has much to do with the experience that our cytologist has had with this technique, and his careful and thoughtful interpretation of the samples. In comparison to cytologic findings obtained by Papanicolaou smears in the diagnosis of endometrial lesions, this technique is far superior. We evaluated this procedure in women who were classically considered to be at high risk for developing endotnetrial abnortnalities, including adenocarcinoma of the endometrium, and have shown that it is a simple
FN = false negative.
Table IV. False positive, false negative, and predictive values with varying prevalence
False Prevalence positives (%) 11100 1150 1120 1/10 l/5 112
80.2 66.8 43.8 27.0 14.1 3.9
False negatives (%)
Predictive value of a positive result(%)
Predictive value of a negative result(%)
0.028 0.058 0.149 0.316 0.708 2.774
25 40 63 80 90 97
100 100 100 100 99 98
screening test, as it should be. However, a screening test must also effectively divide a population into two groups, 10 one comprised mostly, if not entirely, of individuals who are likely to have disease, and the other comprised mostly, if not entirely, of individuals who are likely to be free of disease. We examined the predictive value of this test in terms of two questions. How accurately does the test result predict the presence or absence of disease? If there is an abnormal result, what are the chances that it indicates disease? The predictive value of a test is determined by the complex interaction of three variables: the incidence of false negative results in patients with disease, the incidence of false positive results in subjects without disease, and the prevalence of the disease itself in the group examined. Sensitivity can be defined as the incidence of true positive results in patients with disease. The sensitivity of the Mi-Mark technique was 69/71, or 97%. Specificity can be defined as the incidence of true negative results in subjects without disease. In our study, specificity was 74177, or 96%. Galen and Gambus 11 defined the predictive value as the percentage of positive results that are true positive when the test is applied to a population comprised of both healthy and diseased individuals. We established that the Mi-Mark helix had a predictive value of 97% (Table III).
288
Cramer and Osborne
There were two instances of false negative results, but a review of these two cases yielded no explanation. Possibly, the area of endometrial surface affected was so small that it was missed by the helix. Moreover, false negative results can occur if the cavity is larger than 4 inches, as was the case in one of the two instances of false negatives in our study. The results of the present study are not in total agreement with those of a recent study 9 which concluded that, "It is doubtful whether this technique is suitable for widespread screening and monitoring hormone therapy .... "The sample size in the present study was small, and this, of itself, precludes recommending this procedure as a widespread screening method; but the results are encouraging. On the basis of our test results and varying prevalence rates, the false negative and false positive incidences shown in
Ft>bruan ! , l ~H11
Am . .J. Obstet. Gvnecol.
Table IV can be predicted. We can relv upon a negative result, which means that the patient can be J-cas'iurcd and not subjected to further studv. Is the :VIi-Mark technique like a Papanicolaou test:' That is. does the predictive value of a positive result improve with two or more samplings? \Ve arc presently carrying out further studies specifically in this area to answer this question. ~1ore experience \Vith larger ntunbers of "'-on1en rnay show that this is, indeed, an ideal screening test for the group of women at high risk for endometrial hyperplasia and carcinoma. However. we cannot presently advocate this method as a general screening procedure for women at low risk. \Ve \vish to thank Dr. R. S. Bush, Director, ()ntario Cancer Institute, for his assistance in the preparation of this article.
REFERENCES l. Milan, A. R., and Markley, R. L.: Endometrial cytology by
2. 3. 4. 5.
6.
a new technique, Obstet. Gynecol. 42:469, 1973. Milan, A. R., Markley, R. L., eta!.: Endometrial cytology using the Milan-Markley technique, Obstet. Gynecol. 42:lll, 1976. Montanavi, G. C., Marconato, A., Grismondi, G. L., eta!.: Endometrial lavage as an aid in the cytochemical detection of adenocarcinoma, Cancer 19:1578, 1966. Abramson, D., and Driscoll, S.: Endometrial aspiration biopsy, Obstet. Gynecol. 27:381, 1966. Lubbers, ]. A.: Diagnostic suction curettage without anaesthesia. A.n investigation into the practical usefulness of the Vabra aspirator, Acta Obstet. Gynecol. Scand. (Suppl) 62:1, 1977. Gravlee, L. C.: Jet irrigation method for the diagnosis of endometrial adenocarcinoma, Obstet. Gynecol. 34: 168. 1969.
7. Boschaunn, H. W.: Cytometry on normal and abnormal endometrial cells, Acta Cytol. (Baltimore) 2:520. 1958. 8. Markley, R. L., and Milan, A. R.: Simplified endometrial testing by the Milan-Markley technique, South. Med. ]. 72:452, 1979. 9. Crow, .J., Gordon, H., and Hudson, E.: An assessment of the Mi-Mark endometrial sampling technique, .J. Clin. Pathol. 33:72, 1980. 10. Henderson. M.: Validity of screening, Cancer 37:573, 1976. II. Galen, R. S., and Gam bus, S. R.: Beyond Normality-The Predictive Value and Efficiency of Medica! Diagnoses. New York, 1975, John Wiley and Sons. 12. Gusberg, S. B., and Kaplan, A. L.: Precursors of corpus cancer. IV. Adenomatous hyperplasia as Stage 0 carcinoma of the endometrium, AM. j. 0BSTET. GYNECOL. 87:662, 1963.