Adenocarcinoma of the cervix

Adenocarcinoma of the cervix I. Clinical evaluation and pathologic features HUGH M. SHINGLETON, M.D. HAZEL GORE, M.B., B.S. DOROTHY H. BRADLEY, B.S. SENG-JAW SOONG, PH.D. Birmingham, Alabama A series of 137 patients with primary adenocarcinoma of the cervix is presented with emphasis on clinical and histologic features. Adenocarcinoma has several distinct histologic patterns that may occur in a pure form but often occur (44%) in combinations of two or more patterns. Mixtures of adenocarcinoma and squamous cell carcinoma occur in one third of the patients and squamous intraepithelial neoplasia occurs in one tenth. Transmission electron microscopy is helpful in classifying poorly differentiated tumors. Preliminary recurrence/survival data suggest that the identification of patterns and subpatterns has no prognostic significance, nor is there evidence that adenocarcinoma (or adenosquamous carcinoma) metastasizes earlier, metastasizes more frequently, or is more radioresistant. For most patients, treatment may be the same as that for squamous cell carcinoma but a few selected patients with bulky central lesions may benefit from combined therapy consisting of whole-pelvis irradiation and conservative hysterectomy. (AM. J. 0BSTET. GYNECOL. 139:799, 1981.)

on adenocarcinoma of the cervix presents certain problems for the unwary reader. Many clinically oriented series make no attempt to review or classify the histologic patterns, referring only to "adenocarcinoma."' On the other hand, some papers are oriented toward pathologic features and outline a variety of histopathologic classifications, no two alike and some containing as many as 11 subclasses. 2- 4 Rarely are histologic patterns related to treatment or prognosis. Curability has sometimes been related to grade of tumor 5 ; muciferous tumors resembling those of the bowel are said to have a "poor" prognosis 2 and the endometrial type of tumor supposedly has the "best" THE LITERATURE

From the Departments of Obstetrics arul Gynecology, Pathology, arul Biostatistics, The University of Alabama in Birmingham. Supported in part by Gynecologic Oncology Grant CA 12484, Gynecologic Cancer Education Grant CA 24118, arul Clinical Cancer Education Grant CA 17965. Presented by invitation at the First Combined Annual Meeting of The American Association of Obstetricians arul Gynecologists and the American Gynecological Society, Hot Springs, Virginia, September 3-6, 1980. Reprint requests: Dr. Hugh M. Shingleton, Department Obstetrics arul Gynecology, University of Alabama, Birmingham, Alabama 35294. 0002-9378/81/070799+ 16$01.60/0 © 1981 The C. V. Mosby Co.

of

prognosis. 6 Many studies are small3 • 6 • 7 and observations are often difficult to confirm because of lack of review of pathologic material and lack of uniformity in nomenclature. In some studies comments are limited to pathologic observations with no consideration of clinical implications/ or pathologic findings are related to survival without adjusting for stage of disease or method of treatment. 2 Some oncologists have suggested that there is an "earlier" spread of adenocarcinoma but surgical data to prove this are generally lacking because of the rarity of the lesions. Indeed, even the use of matched squamous cell carcinoma controls is unusual. Not always are adenosquamou~ lesions separately identified. Some consider that adenosquamous carcinoma has a worse prognosis than "pure" adenocarcinoma,2· 8• 9 although this has not been clearly established. Since this institution has had a large number of patients within a limited period of time who were evaluated and treated in a consistent manner, this study was begun to review and classify the pathologic patterns and to correlate them with the clinical features. Furthermore, the availability of major surgical specimens in over half of the patients provided information about nodal spread. Matched control subjects 799

800 Shingleton et al. Am.

J.

April 1. l9HI Obstet. Gvnerol.

Table I. Age, race, and parity of 1,390 women with cervical cancer Age (yr)

No. of patients

Diagnosis Squamous cell carcinoma

1222

Adenocarcinoma and variants

137

Other malignancy of cervix

31

Median Mean Median Mean Median Mean

Table II. Adenocarcinoma compared to all cervical cancers in two time periods Cervical adenocarcinoma

Time period

All cervical cancers

1974-1978

687

51

1979-1980

233

43

No.

I

%

Comparison

7.4>

p < 0.0001

18.5

were selected from a large parallel series of patients with squamous cell carcinoma.

Material and methods Between August, 1969, and july, 1980, 129 patients with a diagnosis of invasive cervical adenocarcinoma were entered in the Gynecologic Tumor Registry. The earlier date was selected since it coincided with the arrival of the first author at the University of Alabama in Birmingham. An additional eight patients were included from the time immediately prior to the larger group since these patients (and all others in the series treated with irradiation) were treated under the direction of the same radiation oncologist. Of this group of 137 patients, 17 were seen for recurrent disease. While their histopathologic findings were included in the tumor pattern component of the study, the remaining tables deal with 120 patients who had primary therapy delivered solely or in part at this institution. Histopathologic slides from initial diagnostic material and any subsequent surgical specimens were reviewed jointly by two of the authors (H. M. S. and H. G.). A control group of squamous cell carcinomas was identified by the biostatistician from the overall data set; this group was matched for age, race, stage, and treatment. Clinical and pathologic data were obtained on these control patients for use in this study and the subsequent clinical publication. The control series was used herein to provide comparative data on parity, degree of differentiation, nodal spread, and recurrence/ survival. Careful attention was paid to differentiating en-

Nulliparous(%)

52 52 47 48 55 53

5.9 13.9

Parity Median Mean Median Mean

Race(%) 3 4.0 2 2.8

White Black White Black White Black

59 41 68 32 62 38

docervical from endometrial tumors: ( l) In 65 of the 120 study patients, the entire uterus was available for study to confirm the diagnosis. (2) In fi.ve patients cervical stump carcinoma had developed many years after the removal of the corpus and the histologic appearance of the tumor suggested a cervical primary. (3) ln many instances, a normal-sized corpus was palpated or demonstrated by ultrasound scanning, or fractional curettage revealed normal endometrium and the cervix was the site of the mass. (4) In instances where doubt occurred with tumor morphologically of the endometrial type, the endometrium was considered to be the primary site and the patient was excluded from this study. All patients selected had tumors with recognizable adenocarcinomatous elements and no patient was included who had a squamous cell carcinoma of the socalled "mucoepidermoid" variety. While special staining (Alcian blue and periodic acid-Schiff) was used at times to establish mucin in some poorly differentiated tumors, routine screening for individual cell mucin was not done as has been advocated by others.~

Results Table I lists the comparative age, parity, and race in 1,390 women with cervical cancer treated on the Gynecologic Oncology Service during the study years. The overall incidence of adenocarcinoma in the study group was 9.3%. Patients with adenocarcinoma were slightly younger than those with squamous cell carcinoma and they had an increased incidence of nulliparity as well as a slightly lower mean parity. Over two thirds of the adenocarcinoma patients were white whereas the comparable figure for the squamous cell cancer patients was 59%. Three patients were pregnant at the time of diagnosis of adenocarcinoma. The total number of patients with cervical cancer ranged from 42 in the first study year ( 1969) to 148 in the last complete study year ( 1979). From 1969 to 1978, the percentage of adenocarcinoma ranged between 3.5 and 7.7 (except for 1973, when the percentage was 13.9) with an overall mean of 8.2. In 1979 and 1980, a remarkable increase in adenocarcinoma occurred (Ta-

Adenocarcinoma of cervix. I 801

Volume 1:19 :"Jumber 7

Table III. Adenocarcinoma of the cervix in previously untreated patients: Age related to clinical stage Clinical Stage Age group

I

10-19 20-29 30-39 40-49 50-59 60-69 70-79

14 27 i8 16 10 6

II

Total III

IV

No.

%

I

3

0.8 12.5 24.2 18.3 21.6 11.7 8.3 2.5

120

100.0

15 29

2

22 26 14

:=:80

I

9 2 3 2

Total

92 (76.7%)

20 (16.7%)

ble II). In these two years, 18.6% of the cervical cancer patients had adenocarcinoma whereas for the preceding 5 years the percentage was 7.2. The clinical stage of cervical adenocarcinoma is related to patient's age in Table III. Of the 120 patients,

lU

6 (5.0%)

2 (1.7%)

Table IV. Adenocarcinoma of the cervix: Treatment related to clinical stage Clinical Stage Treatment

I

92 (77%) had clinical Stage I disease. The median age

Radiation

31

was 47 years; one patient under age 20 was a diethylstilbestrol-exposed patient with a clear cell carcinoma of the cervix. Patients with Stage II to IV disease were significantly older than those with Stage I disease (P = 0.003). The average age for patients with Stage II to IV disease was 55, while the average age of Stage I patients was 45. No significant age difference was found when the adenocarcinoma patients were compared to the patients with adenosquamous lesions. Almost two thirds of the Stage I patients and 54% of all patients \vere treated by total or radical hysterectomy with or without adjunctive irradiation (Table IV). Radical hysterectomy with pelvic node dissection was used as the sole treatment in 44 patients (36. 7%). Eight patients had whole-pelvis irradiation (usually 5,000 to 6,000 rads in 5 to 6 weeks) following hysterectomy after identification of metastatic tumor in pelvic nodes. One patient (Stage IV) received chemotherapy only, and another Stage IV patient refused any therapy. Radiation therapy usually consisted of combined, high-energy. external beam radiation plus intracavitary cesium systems with a goal of delivering 7,000 rads to Point A and 6,000 rads to Point B in 5 to 6 weeks. Fifty-three patients ( 44.2%) \vere treated in this man~ ner. A smaller number ( 13 patients, 10.8%) was treated by conventional total hysterectomy (vaginal or abdominal) preceded or followed by whole-pelvis irradiation, 5,000 to 6,000 rads in 5 to 6 weeks, sometimes combined with vaginal vault irradiation with ovoids or molds used to supplement the central pelvic dosage. Often this was a consequence of lack of a cancer diagnosis preoperatively. In ll of these cases only the radiation component was given at this institution; they

Radical hysterectomy Radical hysterectomy plus radiation* Hysterectomy plus radiation t Chemotherapy No treatment (refused)

43

Total

Total

I%

III!mlw

No.

6 0

0 0

53 44

44.2

0 0

8 13

6.7 10.8

16

7

I

II

2

0 0

0 0

0 0

0 0

92

20

6

36.7

0.8 0.8 2

120

100.0

*Whole-pelvis irradiation following surgery where positive pelvic nodes were encountered (4,000 to 6,000 rads in 5 to 6 weeks). tTotal vaginal or abdominal hysterectomy preceded or fol-

lowed by whole-pelvis irradiation.

were included in the series since (1) the surgical specimen was available in all cases for review, (2) the size and thus the clinical stage of the lesions were available, and (3) follow-up information was available. In only two instances has this combined approach been utilized at this institution since the surgical candidates with Stage I lesions ordinarily had radical hysterectomy and pelvic node dissection, and generally our radiation therapists did not in principle believe that cervical adenocarcinemas lvere more radioresistant than squamous cell carcinomas, thus obviating the combined approach. When the clinical information was abstracted the type and size of lesion, especially in the Stage I and Stage II lesions, were noted. Overall, lesions were endophytic in 32%, exophytic in 40%, ulcerative in 19%, and polypoid in 9%. That two thirds of the patients had lesions other than endophytic was somewhat surprising. When the lesion type was related to size, the larger lesions were found to be the exophytic ones, two thirds

802 Shingleton et a!. Am.

J.

April I. l!JHI Obstet. G; new I.

Table V. Size oflesion related to node metastases: Stage I Adenocarcinoma

I

Pelvic unilateral

Size

No. of patients

~~em

2cm ;;:,3cm

24 14 16

3

Total

54

9

~

0 6

Pelvic bilateral 0 0 l

Squamous cell carcinoma

I

Aortic 0 0 l

I

%

No.

of patients

I

Pelvic unilateral

I

Pelvic bilateral

I

%

0 21.4 50.0

28 16 8

1 4 0

2 0 0

10.7 23.0 0

20.4*

52

5

2

13.5*

*Differences not statistically significant.

of these being 3 em or greater, whereas only about one seventh of all the lesions were this large. Of special interest with regard to proposed combined irradiation and conservative hysterectomy as treatment of bulky Stage IB and barrel-shaped lesions,10 14 of 22 patients with Stage lB lesions measuring 3 em or more were treated by radical hysterectomy and pelvic node dissection. Of this surgical group, eight have experienced recurrence (six in the pelvis and two with distant metastases); six of these eight patients died of disease. The remaining eight patients were treated with radiation therapy and to date only one has developed a recurrence (in the pelvis). Only eight patients with barrel-shaped lesions were encountered in the adenocarcinoma study group. One of the eight had clinical Stage IB disease, six had Stage liB, and one had Stage II I B. All of these patients were treated with radiation. To date, three have developed recurrences at 2, 5, and 27 months. Exenterative surgery has presumably saved all of these patients. Six of the eight barrel-shaped lesions were adenosquamous in type. Size of the original lesion was correlated with the pelvic node status at operation in 54 Stage I patients with adenocarcinoma and matched control subjects with squamous cell carcinoma (Table V). The I em or less category included the same percentage of patients in each pathologic category whose diagnosis was established by conization of the cervix; some patients had small visible lesions. In this overall group, no positive nodes were encountered in the adenocarcinoma patients but 10.7% of the control subjects with squamous cell carcinoma had positive nodes. The percentage of positive pelvic nodes encountered in the 2 em group was almost identical for the two tissue types, whereas in the 3 em or larger lesions, 50% positive nodes were found in the adenocarcinoma group. The matched control subjects had no positive nodes in this siz.e category, most likely a sampling bias, since in a larger sample of squamous cell carcinomas in this size category, four of 17 (23.5%) had positive nodes. Overall, 20.4%

of the Stage I adenocarcinoma patients had posltlve nodes whereas 13.5% of the Stage I squamous cell carcinoma patients had positive nodes; this difference was not statistically significant. Only one patient (clinical Stage I adenocarcinoma) had a positive aortic node at the time of exploratory laparotomy for radical hysterectomy. Radiation treatment was given to the pelvis and para-aortic region, but the patient died of disease.

Pathology studies Histologic patterns. Review of the pathologic material of the 137 patients consisted of an initial sorting into two major patterns (adenocarcinoma and adenosquamous carcinoma) and further classification into seven subpatterns. The msot common subpattern was the endocervical type. which consisted of malignant endocervical glands with or without focal intracellular mucin. The second most common subpattern was papillary, which did not occur as a pure subpattern but was commonly associated focally with other subpatterns. Mucinous tumors, i.e., mucin in glands or stroma (resulting in a myxomatous pattern), was seen as a pure subpattern II times and was mixed with other subpatterns in an additional 11 instances. The "myxoma cervicis" pattern was seen eight times. In the adenasquamous category, endocervical and mucinous subpatterns were the most common (Fig. 1). The adenoid cystic subpattern (Fig. 2) was seen in a total of 21 patients. This pattern was usually f(xal: even among the I 0 patients in whom it was the only subpattern described, it appeared as f(>Cal differentiation within a less-differentiated tumor. More than half of the patients with this subpattern had adenosquamous tumors. The endometrioid pattern reported by other authors was not encountered in this series probably because our selection process excluded most lesions with an endometrioid appearance from consideration. Furthermore, some of the lesions interpreted by us as endocervical might be called endometrioid by other authors. Probable squamous elements within areas of poorly

Adenocarcinoma of cervix. I 803

Volume 13'1 Number 7

differentiated adenocarcinoma were found in 15 patients during the review. Seven of these patients had tissue available for electron microscopic study; squamous elements were specifically identified in the tumors by this method and the patients were reassigned to the adenosquamous category. This confirmation of the mixed nature of the tumors in all seven of these patients as well as the similarity in grade of tumor in the 15 "probable" squamous category patients caused us to combine the "probable" malignant squamous element patients with the adenosquamous carcinoma patients in Tables VI and VII. Ferenczy," who studied 18 undifferentiated tumors, found that 1-! were squamous cell, two were glandular, and two were undifferentiated. This also emphasizes the high probability that undifferentiated tumors or such areas within cervical carcinomas will be of the squamous cell type. 1'\o patients had the "glassy cell" type of lesion as described by Glucksmann and Cherry 12 and others. 13 Two patients with adenosquamous lesions had complete separation of the malignant squamous cell carcinoma and adenocarcinoma elements and these were considered to be "collision" tumors. In all other instances. in the adenosquamous category a spectrum of admixtures of the malignant elements was observed. The adenosquamous lesions were predominantly grades 2 and 3 and were different in relation to grade frorn the adenocarcinoma group at a highly significant level (Table VI). No statistically significant difference in incidence of positive pelvic nodes was found between adenocarcinoma and adenosquamous carcinoma in spite of the previously observed more advanced grade in the latter lesions (Table VII). Although the less-differentiated lesions (grades 2 and 3) were associated with a higher percentage of positive nodes among patients with clinical Stages I and II adenocarcinoma, the difference between these and grade l lesions was not significant, perhaps because of the small numbers in the grade 1 category (Table VIII), nor was the difference significant if grades I and 2 were compared to grade 3. Comparison with the matched control subjects revealed that grade I squamous cell carcinoma was associated with 9.1% positive nodes and grades 2 and 3 had 16.7% positive nodes. This difference was not significant, nor did the squamous control subjects differ significantly from the adenocarcinoma group in overall incidence of positive nodes. The suggestion by Gusberg and associates 14 that stroinal response could be related to node rnetastasis

was not confirmed in this study. Patients with an excellent or good lymphocytic response within the tumor

I

ADENOCARCINOiviA (N:::92)

30 25

>U 20

z

5LLJ 15

~ [1 L i

~ 10

EC

PURE

~

MIXED I OTHERWITH PATTERN

•

MIXED WITH 2 OR MORE PATTERNS

nL

5

O

D

PAP

ll cl. cf. MUC

A/C

HOB

~

~ ~

AD/AC

?SO

ADENOSQUAMOUS CARCINOMA (N=45)

i~~~~

8 10 0::

,.• rfAI n_r.

5

~

0 11 6

EC

PAP

11%:11 I MUC

CJ

A/C

HOB

Fig. I. Histologic patterns and subpatterns in 137 adenocarcinomas. EC = Endocervical; Pap = papillary; Muc = mucinous: A/C =adenoid cystic Hob= hobnail/dear; Ad/ Ac = adenoacanthoma; ?Sq. = probable malignant squamous

element. Table VI. Adenocarcinoma of the cervix: Cell type related to histologic grade, Stage I

IAdenocarcinoma _I varied patterns

I

Adenosquamtras* lesions

I

Hi>tologic No. of %of No. of %of patients patients patients patients grade 21

44

3

2,3

27

56

41

Total

48

I Comparison

'>p

7.

< 0.0001

93

44

*Grade of adenocarcinoma considered for this comparison. stroma (N = 45) had a 20% incidence of positive lymph nodes. Those with a poor response (N = 17) had only a 12% incidence of positive nodes, exactly the opposite of what might be predicted. Ultrastructural studies. A total of 36 specimens exhibiting varied histologic patterns- was available from 27 patients with adenocarcinoma. In seven of these patients, material was available prior to and during radiation therapy. Fig. 3 through 7 demonstrate certain light microscopic and ultrastructural features of these tumors. In the mucin-containing tumors, there was ordinarily a decrease in (and often an absence of) mucin production as the tumors became more dedit~ ferentiated. Ultrastructural studies were found to be

804

Shingleton et al.

April I. I'll" Am. J Obstet. Gvnecol

Fig. 2. Adenoid cystic carcinoma patterns.A, Classic adenoid cystic pattern. Lobulated groups of cells with seamy cytoplasm and small uniform nuclei surround glandlike spaces. (Hematoxylin and eosin. Original magnification x 350.) B, Cylindromatous pattern of adenoid cystic carcinoma with the classic central eosinophilic masses associated with cylinders. (Hematoxylin and eosin. Original magnification X 200.) C, Similar overall pattern but with clear cytoplasm-part of a tumor which included less well-differentiated foci of dear cell carcinoma . (Hematoxylin and eosin. Original magnification X 350.) D, A focus of an adenoid cystic pattern within stroma beneath a surface layer of cervical intraepithelial neoplasia which appeared to be an independent lesion. (Hematoxvlin and eosin. Original magnification X 350.)

Volume 1:>9

Adenocarcinoma of cervix. I 805

Number 7

Fig. 3.A andB, Mucin production in an adenocarcinoma of the myxomatous pattern. The clear areas represent non-membrane-bound mucin (M) including some in stroma (lower right) while some membrane-bound mucin droplets (arrows) are near the gland lumen. Microvilli are seen at the luminal surfaces (L) of the cells. (A, Original magnification x 5,400. B, Original magnification X 15,000.)

806

Shingleton et al.

\pnl I !'lXI Am .. J. Ob,!cL <.vnnol

Fig. 4. Poorly differentiated adenocarcinoma. Mucogenesis is absent. Dilated cisternae of smoothsurfaced endoplasmic reticulum are abundant. Nuclear borders are deeply indented. Note narrow lumina (arrows). S =Stroma. (X 5,400.)

particularly helpful in categorizing patients with grade 3 lesions. The fine structural features of squamous cells, especially cytoplasmic tonofilaments, filament desmosome complexes, and intercellular nticrovilli, allowed identification of squamous cell components even in poorly differentiated tumors. Of special interest were surfaces of adenocarcinoma cells bordering on gland lumina (Fig. 7). Often these lumina were so small that they were not appreciated by light microscopy (Fig. 4). The surface microvilli were at times similar to those of normal columnar epithelial cells but usually were elongated, irregular, clubbed, or blunted. No cilia

were identified in any of the tumors, even in the welldifferentiated ones. Several specimens had ceiis with filaments entered by altered microvillus-like structures. but cross sections established that they vvere nticro\'illi, not cilia (Fig. 7, A and B). Such prominent filaments have not been observed in relation to surface microvilli in normal columnar epithelial cells. Observations on patients with material obtained before and during radiation therapy failed to reveal any major identifiable changes other than swelling and disruption of mitochondria and cytoplasmic vacuolization. In two patients with mixed adenosquamous lesions. the

Adenocarcinoma of cervix. I 807

Volume U9 Number 7

Fig. 5. A clear cell area in an adenosquamous carcinoma. Abundant material, possibly glycogen, washed out in processing, distends the cells, displacing the other cytoplasmic organelles to the periphery. ( x 5,800.)

Table VII. Adenocarcinoma of the cervix: Positive pelvic nodes related to cell type, Stage I Positive pelvic nodes No.

of patients

No.

I

%

Adenocarcinoma-varied patterns

30

4

13.3>

Adenosquamous lesions

27

6

22.2

Total

57

lO

17.5

squamous tonofilaments appeared to be accentuated in sequential biopsies as radiation dosage accumulated. The significance of this observation is unknown. Preliminary observations concerning recurrence and survival

Another publication is planned to report survival and recurrence in relation to treatment and cell types after

ComparisMt Not significant

more time for follow-up elapses. Certain trends, however, appear to be evident now. When compared to the matched control subjects (Fig. 8), the adenocarcinoma patients do not differ at this time from their control subjects in relation to 3-year survival. The adenocarcinoma and adenosquamous carcinoma groups do not appear to differ when corrected for stage (Fig. 9). Size within Stage I and the stage itself appear to be significant

808

Shingleton et al. Am.

J.

.\pril I. 191'{1 Ohstet. G1 necol

Fig. 6. Light and electron microscopic features of a poorly differentiated adenosquamous carcinoma, demonstrating admixtures of cell types. A, There is focal differentiation into a glandular component. (Hematoxylin and eosin. Original magnification x 250.) B, The cell at the left is a well-differentiated neoplastic squamous cell with coarse tonofibrils (S). There is also a mucin-producing adenocarcinoma cell (A) and an undifferentiated cell (U). (Original magnification x 12,420.)

prognostic indicators; lesions 2 em or less appear to have a good prognosis v:hereas lesions 3 em or more in Stage I appear to parallel the curve for Stage I I lesions (Fig. l 0). In addition to these observations, the various subpatterns as listed in Fig. I were tested; at this time, neither good nor bad prognosis groups could be identified.

Comment "'ALdenocarcinoma of the cervix rnay be increasing in incidence. 15 Some authors have noted a gradual rise in recent years whereas we recently found a dramatic increase. In some reports, the increase may be a relative one in relation to the numbers of squamous cell carci-

Adenocarcinoma of cervix. I 809

Volume J:l9 Number 7

Fig. 7. Luminal surface features of neoplastic glandular cells. A, Abundant cytoplasmic tonofibrils and filaments (arrows) are prominent at luminal surfaces, entering the microvillus-like structures. (X 23,000.) B, High-power view demonstrating microfilamentous substructure in cross section. (X50,000.)

nomas encountered, but that does not seem to be the explanation at this institution. Indeed, the sudden increase led to our selecting this subject for study. Recently. the suggestion has been made that oral contraceptives may have some etiologic significance in relation to adenocarcinoma but the arguments are speculative. 16 Because of the known ability of the endocervical cells to respond with a "pill pattern" or "pregnancy pattern" which is sometimes very difficult to distinguish from adenocarcinoma, further studies of a possible relationship are needed. Unfortunately, data regarding oral contraceptive use are lacking in our study. The preliminary analysis of tumor recurrence and patient survival in this series suggests that surgery and radiation are both effective forms of treatment. In regard to selection of treatment, it appears that radiation is preferable to radical surgery in bulky Stage IB lesions since, in spite of removal of these lesions and adjunctive postoperative pelvic radiation in several instances, the recurrence rate was quite high. Perhaps a better approach in these patients is to deliver wholepelvis radiation prior to a conservative hysterectomy, as suggested by others. 10 Surprisingly, however, in the few

Table VIII. Adenocarcinoma of the cervix: Positive pelvic nodes related to histologic grade, Stages I and II

Histologic grade

Positive pelvic nodes No. of patients

No.

17

I

%

5.9>

2,3

45

10

22.2

Total

62

11

17.7

Comparison Not significant

(N = 8) patients with the same size bulky Stage IB lesion (3 em or more) treated by radiation alone, only one of the eight has developed a recurrence to date, raising the question of what the adjunctive hysterectomy would have added. While the gynecologic oncologists at this institution have had no objection to a planned combined radiation and conventional hysterectomy approach to remove the central disease and potentially to prevent central pelvic recurrence, there has been little opportunity to implement this plan up to the present time. The seven barrel-shaped lesions and the

810

Shingleton et al.

April I, llJH I Am . .f. Obsttt. (;vnecol.

100

Squamous Cell Carcinoma ----"'Stage 1 r-

Adenoca'rcinomo Stage I

30 f'f:!f! Ufl.'-S

2~r

~

------ 1\[J[ l'tU(AHLINOMA '.OUAI'lJU~ U l L LAIKINUMA

-o-

IU[

-e-

\flllA,IlJ)((lL lAI
II- [11

~

5-YI<',IJi
CJl

84

90 iS

87 ~·~

"

')~

0~--7---~2--~3~--471--~5--~6~~7~--~8--~---J~~L-~. II 9 10 12 I

I

I

I

Years from Date of Diagnosis

Fig. 8. Adenocarcinoma and matched controls. Survival curves by clinical stage. 100

--J~-

.._ .....JL- . . J . . I _ - - - - -l.....J ..I.. ..J

Adenocarcinoma, Stage I

0>

L~ :~~ ~

l 40~u c:

~

rf

30

~-~\-----------.1......1

Adenosquamous Carcinoma Stages n,m,Ill

__ .... ________ .... __ ...1 . Adenocarcrnoma, Stages U.,m,N

20

-o--+-

10

----

-6-

0

2

!

~ ADENOCARCINOt'o\ ADEMOSQIJA~US

CARCINOMA

ADENOCARC INOM AOCNOSQ~US

CARCI NOM

STAGE I I II-IV II-IV

3-YRSURVIVAL ~- ~

58

85

l3

Rl

16 12

38

<•2

10

4 9 5 6 7 6 Years from Dote of Diagnosis

ll

12

Fig. 9. Adenocarcinoma and adenosquamous carcinoma. Survival curves by clinical stage.

eight bulky (3 em or more) Stage I lesions which qualified for this method represented only 13% of the patients treated in the series. Even without the combined treatment, however, 12 of these 15 patients are living without evidence of disease at varying lengths of time. which leaves little room for improvement and perhaps supports the view of our radiation oncologists that radiation alone is sufficient treatment. Additional follow-up on all of the patients, however, may give more insight into the effectiveness of various forms of

treatment when adjusted for stage, size, grade. and histologic type. The suggestion by Nogales and Boteiia-Liusia 17 that adenocarcinoma of the cervix has an extremely high rate of nodal metastases in all stages was not confirmed in this series, nor was it noted by Webb and Symmonds'" and Baltzer and associates 19 in other series containing large numbers of patients who had surgical oroCf•dures for adenocarcinoma oft he ct>rvix. In studving adenosquamous carcinoma that metastasized, we '

-

-

-

-

-

-

--

--

-

--

- --

-

-

-

I

Adenocarcinoma of cervix. I 811

Volnnw l:l!l

:\umber 7

r Stage I,

~ 1em.

- - - J...- ...1.-L..

-'-----..L- ..1_ - - - - __ ,

Q)

Vl

0

Q)

Vl

0 Q)

Stage I, ?3 em.

0> 0

--..1

'E 40 ~ .....

cf.

30

-----------..J

20

.........

Stage m/Ill', All Sizes

10

0.0

....-t:r-

-.-

-o-

.....--

0.5

10

SlAG

II III/IV

lESIOK SIZE ~I ut. 2c11. l '"· >All

All

35 3.0 25 2.0 15 Years from Dote of Diagnosis

.!...ill.·

HRDISlA\l

mEmEm

40

91

21 22 19 I

85

40

60 50 <14

45

5.0

Fig. 10. Adenocarcinoma of the cervix. Disease~free curves by clinical stage and lesion size.

also were unable to confirm the observation 12 that the adenocarcinoma component tended to be the dominant pattern in nodal metastases: in all instances in our series, the metastases resembled the primary tumor. Ti1e multiplicity and combinations of subpatterns as reported in this series are unique: presumably other investigators have selected the dominant pattern and have not stressed the presence of multiple patterns. In this series the sub patterns were pure in only 57% of the cases. The adenoid cystic pattern is of interest because it has been reported 20 as a rare variant of cervical cancer that deserves some special niche. Our previous report 21 and additional study during this project demonstrate that these tumors diagnosed by light microscopy are not a!wavs the same as adenoid cystic carcinomas of other parts of the body at the fine structural level. The occurrence of this subpattern at least focally in 21 patients in this series ( 15.4%) suggests that it is relatively common in adenocarcinoma, perhaps with little or no clinical significance. The l~1ct that over one third of the patients in this series had a malignant squamous cell element and 14 additional patients had squamous intraepithelial neoplasia on the surface of the cervix adjacent to the adenocarcinoma is perhaps not surprising in view of the probable derivation of the mixed tumors from undifterentiated, multipotential subcolumnar cells in the endocenix. Julian and associatesH and more recently Maier and ~ orris 22 reported incidences of 19% and

43%, respectively, of squamous intraepithelial neo-

plasia associated with cervical adenocarcinoma. Our percentage association ( 10.2) is perhaps closer to an incidence figure. In spite of the fact that the adenosquamous lesions were less differentiated than those of the adenocarcinoma group, preliminary data do not suggest that these patients have a poorer prognosis; therefore, the relevance of major efforts to identify a squamous cell component such as by transmission electron microscopy is questioned. The finding of similar rates of lymph node metastases also supports the concept that such a classification is of little value. The supposed association of pregnancy with adeno· squamous carcinoma (17%) and adenocarcinoma (5%) reported by Glucksmann and Cherry 12 was not observed in this series. Two of the three pregnant patients in our series had adenosquamous lesions (4.4%); the comparable figure for pregnancy associated with adenocarcinoma was l %. Glucksmann and Cherry also described a glassy cell carcinoma which they considered the most undifierentiated type of mixed adenosquamous tumor with little or no differentiation in a squamous or an adenocarcinoma direction. Littman and associates 13 agreed with the concept but failed to document that these are mixed tumors. In our 45 patients with mixed adenosquamous lesions, we did not find a single example that met the classic criteria described by Glucksmann and Cherry because selection criteria for our study required focal differentiation into an adeno-

812

Shingleton et al. Am.

carcinoma pattern. We did observe in some instances areas suggestive of the "glassy" pattern focally. but not

as a pure pattern. \Ve have had the opportunity to study by electron microscopy at least one tumor that seemed to meet all the criteria but in the material studied only poorly differentiated malignant squamous cells were found. We are thus unable to confirm the mixed natures of such tumors or at this time to add information about "glassy cell" carcinomas. The unexpected finding within the adenocarcinoma group of seven patients who had had previous cryosurgery of the cervix (compared with no such patients in

J.

April I. I'JRI Obstcl. Gmc·• ',(.

the matched control group) is of interest. Whether 1he cryosurgery was used to treat documented cervical in· traepithelial neoplasia or ,,·as f(H "cervicitis" \\ ithoul benefit of diagnostic biopsy is unknown, nor is it kno\1 n whether the treatment contributed to delay in diagnosis of the adenocarcinomas.

We would like to thank Drs.j. ~lax Austin.jr .. and Edward E. Partridge for allowing inclusion of some of their patients in the study, and Mr.John Wi!ters t(lr his technical assistance.

REFERENCES I. Rutledge, F. N., Gutierrez, A. G., and Fletcher, G. H.: Management of stage I and II adenocarcinomas of the

2.

3.

4. 5. 6.

7.

8.

9. IV. "'

II.

uterine cervix on intact uterus, Am. J. Roentgenol. Radium Ther. Nucl. Med. 102:161, 1968. Abell, M. R., and Gosling. J. R. G.: Gland cell carcinoma (adenocarcinoma) of the uterine cervix, AM. j. 0BSTET. GYNECOL. 83:729, 1962. Anderson, M. C., and Fraser, A. C.: Adenocarcinoma of the uterine cervix, a clinical and pathological appraisal, Br. J. Obstet. Gynaecol. 83:320, 1976. Gallup, D. G., and Abell, M. R.: Invasive adenocarcinoma of the uterine cervix, Obstet. Gynecol. 49:596, 1977. Weiner, S., and Wizenberg. M. J .: Treatment of primary adenocarcinoma of the cervix, Cancer 35:1514, 1975. Hurt, W. G., Silverberg, S. G., Frable, W. J.. Belgrad, R., and Crooks, L. D.: Adenocarcinoma of the cervix: Histopathologic and clinical features, AM. J. 0BSTET. GYNECOL. 129:304, 1977. Parker, J. C., Jr., Van Nagel!, J. R., Jr., and Bissig, T.: The histomornholovic snectrum of endocervical (Mullerian) adenoca~ci~o~a--~ potential prognostic indicator, J. Surg. Oncol. 9:267, 1977. Julian, C. G., Daikoku, N.H., and Gillespie, A.: Adenaepidermoid and adenosquamous carcinoma of the uterus, AM. j. 0BSTET. GYNECOL. 128:106, 1977. Wheeless, C. R., Graham, R., and Graham, J. B.: Prognosis and treatment of adenoepidermoid carcinoma of the cervix, Obstet. Gynecol. 35:928, 1970. t~elson, A. j., Fletcher, G. H., and \Vharton, j. T.: Indications for adjunctive conservative extrafascial hysterectomy in selected cases of carcinoma of the uterine cervix, Am. J. Roentgenol. Radium Ther. Nucl. Med. 123:91, 1975. Ferenczy, A.: The ultrastructural morphology of gynecologic neopiasias, Cancer 38:463, i 976.

Discussion DR. J. TAYLOR WHARTON, Houston, Texas. Shingle-

ton and colleagues have meticulously studied 137 patients with adenocarcinoma of the cervix. The report has an immediate impact because the authors were responsible for the treatment planning, correlation of the histologic patterns, and follow-up for 120 of the patients in the study. There are two distinctly identifiable parts of this report. The first is a review of the clinical features, treat-

12. Glucksmann, A., and Cherry, C. P.: Incidence, history and response to radiation of mixed carcinomas (adenoacanthomas) of the uterine cervix, Cancer 9:971. 1956. 13. Littman, P., Clement, P. B., Henriksen, B .. Wang, C. C., Robboy, S. J., Taft, P. D., Ulfelder, H., and Scully, R. E.: Glassy ceil carcinoma of the cervix, Cancer 37:2238, 1976. 14. Gusberg, S. B., Yannopoulos, K., and Cohen, C. J.: Virulence indices and lymph nodes in cancer of the cervix, Am. J. Roentgenol. Radium Ther. Nucl. Med. lll:273, 1971. 15. Davis, J. R., and Moon, L. B.: Increased incidence of adenocarcinoma of the uterine cervix, Obstet. Gyneco!. 45:79, 1975. 16. Czernobilsky, B.. Kessler, I., and Lanat. M.: Cervical adenocarcinoma in a woman on long term contraceptives, Obstet. Gynecol. 43:517, 197 4. 17. Nogales. E., and Botella-Liusia, J.: The frequency of invasion of the lymph nodes in cancer of the uterine cervix, AM. j. 0BSTET. GYNECOL. 93:91, 1965. 18. Webb, M. J., and Symmonds, R. E.: Wertheim hysterectomy: A reappraisal, Obstet. Gynecol. 54:140, 1979. 19. Baltzer, J.. Kopcke, W., and Zander, J.: Das operientc adenokarzinom der cervix uteri, Geburtshilfe Frauenheilkd. 39:1011, 1979. 20. Benitez, E., Rodriguez, H. A., Rodriguez-Cuevas, H., and Chavez, G. B.: Adenoid cystic carcinoma of the uterine cervix: Report of a case and review of four cases, Obstet. Gynccol. 33:1757, 1969. 21. Shingleton, H. M., Lawrence, W. D., and Gore, H.: Cervical carcinoma with adenoid cystic pattern, Cancer 40: 1112, 1977. 22. Maier, R. C., and Norris, H. S.: Coexistence of cervical intraepithelial neoplasia with primary adenocarcinoma of the endocervix, Obstet. Gynecol. 56:361, 1980.

ment, and results. The second is a pathology of the histologic patterns and correlation with the clinical data. The design of the study is such that a number of questions can be answered. The selection of certain pretreatment patient characteristics, such as age, race, and parity, and then tumor characteristics, such as stage, degree of ditTerentiation, and nodal spread, allowed the use of a control group of appropriately matched squamous cell carcinomas.