The lymphatic dissemination in endometrial A study of 188 necropsies
ERLE HENRIKSEN. M.D. Los Angeles, California
To more clearly define the biologic artivity of mdomrtrial carcinoma, 188 mses .\elected at random were studied at necropsy. In 51 per ern! of the casts death was attributed to thr endometrial carcinoma. Observations demonstrate that dissemination is mort scattered and lrss predictable than noted in cervical rancer; that the biologic potential is not invariably based on the site of the primary lesion, the depth of invasion, or thl' histologic differentiation. Difficulties in pnforming a thorough rxarnination of the lymphatic system in this series include the increased age of the patients and a higher incidence of medical disPascs and surgical procedures. Tht reported data illustrate the erratic natural history of the disease but do not permit the factual incidnzce of node or organ involvement. However, the potential viciou.mess of the disease is rlearlv drmonstrated.
THE ORIGINAL concept of this study. now admittedly both naive and presumptuous, was to attempt to determine the involvement of the lymphatic system in the dissemination of endometrial carcinoma. In its organization, the study has gone through three epochal phases. The first phase limited the study to the involvement of the lymphatic system, regardless of the gross character of the disease. The second phase was the awareness that the site of the primary lesion and the histologic grade were probably important and should be an integral part of the investigation. The third and last phase. following the tabulation of the data, was the realization that many preconceived ideas were not supported by this study. It is evident that the data. based on random selection, neither reveal nor permit definitive correlation, but do demonstrate the numerous pathways of dissemination and the possible sites of distant metastases.
data must be interpreted accordingly. Fiftv-two of the necropsies were either performed or supervised by the author. The remaining 136 cases were selected from a series of over 700 necropsy protocols collected from many sources over a period of 35 years. The series is divided according to the stage of the disease as established at surgery or necropsy. The stage of the treated cases was based on pretherapy assessment (Table II). Additionally, the histologic grade of differentiation was examined. The treated and the untreated cases were studied separately (Tables III and IV). Seventy (42 per cent) cases received various forms of treatment, in retrospect often inadequate and occasionally mappropriate by today's standards.
Ciinicai staging Though the system of staging employed in this report was originally designed to provide a dependable surgicopathologic system, its lack of value is apparent. At least it may afford insight into some of the difficulties encountered in a study of this type. To establish the probable stage in each case the extirpated uterus was opened at surgery or necropsv and examined grossly. The stage of the treated cases was based on the clinical estimate prior to therapy. It is evident that this system of classification is certainly not reliable. A Stage I, Grade l case, later found to have nodal and/or organ involvement, should more properly be considered a Stage IV case. Staging based on the size of the uterus disregards the possible presence
Material A total of 188 cases of endometrial carcinoma were studied at necropsy (Table I). Sinc:e the series does not
represent a group of consecutive cases the reported
From The Hospital uf the Guod Samaritan. Presented at the Ninety-eighth Annual Meeting of the American Gynecological Society, Coronado, Cali:forniaApril9-12, 1975. Reprint requests: Dr. Erie Hfnriksen, 1136 W. Sixth St., Los Angeles, California 90017.
570
\'olume 1~:1 :\umber 1;
Lymphatic dissemination in endometrial carcinoma
571
Table I. Sites of metastases in 79 treated and l 09 untreated cases
79 109
Treated Untreated
62 61
62 57
18 13
32 27
22 24
30 36
33 34
Distant metastases
Cancer deaths
(%)
(%)
65 61
52
50
Distant metastases Cancer deaths (%)
(%)
5 31 35 38
49 54 59 68
27 31 65 56
32 48 67 56
56 68 80 89
56 64
Stage I 4 I
37 26 17 34
II
4
III
I 4
2':>
IV
I
15
4
25 9
39 54 .S9 65 68 76 87 78
43 50 71 71 68 72 60 67
19 23 18 12 8 16 12
11 46 24
35 28
~2
40 57
of fibroids. pyometra. elc.; though often useful, it is not completely reliable. Similarly, palpable changes in the parametrium are not always indicative of malignant induration. The following classification based on the gross examination of the extirpated uterus was used in this study:
Stage I. The disease involves less than one half of the uterine wall and appears confined to the upper half of the corpus. More than one half of the recognized primary sites seem to originate in this area. Stage II. The disease involves the entire corpus but does not extend into the endocervix. Stage l!I. The disease involves the entire corpus and/or the lower segment, including the endocervix. Stage IV. The disease involves the corpus and extends to contiguous organs or has spread distantly. Histologic grading
Observations pertinent to the importance of histologic differentiation are conflicting. In some cases, its importance appears evident; in other cases correlation between the histologic grade and the growth activity of the disease was totally lacking. The diagnosis of adenoacanthoma was made in approximately 10 per cent of the series. The frequency of diagnosis varies with the observer. Those adenocarcinomas with associated cellular patterns consistent with squamous-cell metaplasia were included in the series and no apparent difference in the biologic activity was noted. Eleven cases with both adenomatous and squamous elements, characterized by marked anaplasia. were not included because of inadequate investigations.
3
38 12 32 28 32 33 33
11 35 29 38 40 40 53
56
73
78
Endometrial carcinoma does not lend itself to exact microscopic grading because of the variable cell patterns in different areas of the same tumor. To consolidate the material, Grades I and II are listed as Grade I or as "differentiated"; Grades Ill and IV are listed as Grade IV or "undifferentiated." The lymphatic system (Fig. 1)
Though not predictable, the carcinomatous dissemination in cervical cancer cases tends to follow a more orderly pattern, whereas, in the endometrial carcinoma cases, not only is the pattern less predictable. but the wider spread suggests, at times, an almost explosive propulsion of the malignant emboli in all directions. In the cervical cancer case, it seems that once the motion of the malignant embolus has started. there tends to be a more orderly progression from one node group to another. This orderliness of travel does not occur in all cases, for both bizarre and unexplained sites are not uncommon. On the other hand, it appears that, despite the probable origin of the malignant embolus in endometrial carcinoma, more routes of dissemination are available. An occasional pattern will demonstrate the preconceived ideas based on the primary site, the depth of invasion, and the histologic differentiation. We must bear in mind that it is not possible to report all of the observations. Some are conflicting, some are unexplainable, others are suspect of inadequate or erroneous interpretation. Until definite patterns or systems of investigation are established. conflicting statistics will continue to be generated. Nevertheless, the following observations, based on the gross and microscopic examination of the necropsv material are noted:
572
\(1\l'lllhlT J:~l. \~l/.-1
Henriksen
:\!ll. J.
Tabie HI. Sires of metastases by stage in 94 treated and untreated histologic Grade
()b"ilt'l. (~!llCUd.
case~
Pelvic nodes
Aortics
Aortir.1 onh
Bowel
Bladder
Adnrxa
Parametrium
Distant metastasn
No.
(%)
(%)
(%)
I 'lc!
('lc!
(%)
{ o/c)
{o/c)
{%)
Treated Untreated
13 24
46 33
62
15 21
H I :i
H
f'> )~
31
II
Treated Untreated
6
()7
:1:\
55
67 73
33
II
9
IH
IH
':27
36
III
Treated Untreated
13 12
62
62 60
15
2:i
2:1
2~\
:ll
IV
Treated Untreated
7 H
71 100
Stage
75
~\3
3~\
4~\
57
:lH
6~\
:\3 29
~\H
2~\
4
4
12
:u
~~~~
:·n
."iH 2~!
7.·)
Cancer
nllL\t'
ofdrath
.,._:)
67
15
64
.~4
:l:l
:>K
62 50
57 75
71 HH
71 73
Table IV. Sites of tnetastases by stage In 94 treated and untreated histologic (;rade 4 cases Distant
Cancer
raiL\/'
uf drath { 7r)
12
Treated Untreated
14
67 43
II
Treated Untreated
12 22
Ill
Treated Untreated
10 15
H7
IV
Treated UntreateJ
6
58
42 50
58 50
4~\
H 36
67 64
67
17
67
7:\
I)
6H
GO
60 HO
~\()
40
7
53
67 100
67 6i
I. The original lesion may invade superticiallv or with tonguelike extensions invade the myometrium. 2. Invasion of the blood vessels is probably not uncommon during the transport or extension of the disease through the endometrium and the myometrium. This could explain the distant nodal and organ involvement in the very early stages of the disease. 3. The corpus. based on some of the myometrial invasion patterns. appears to be divided into at least three poorly definable compartments. The borders of the compartments tend to partially overlap. This compartmentalization would account for the route selection in some cases. The overlapping would account for the unpredictable dispersal patterns. 4. The high incidence of peivic surgery and medical diseases. along with the increasing age, probably accounts for the marked visible changes in both the lymph channels and the node groups. That these physiopathologic changes can obstruct the main channels and derange the usual node group patterns is frequently observed. This could be a factor in the wide scattering of involved nodes. 5. A Stage IV. Grade 4 lesion mav extend into both the bladder and the rectum with no demonstrable nodes or other organ involvement.. On the other hand
50 G7
17 t)7
HiO
100
]00
100
a Stage I. Grade I case may have extensive dissemination. 6. Dissemination from the cndoccnix or cervix docs not inevitably follow the pathways usuallv taken bv carcinoma of the cervix. With the lesion in this area it is possible to have high dissemination without umcomitant involvement of the pelvic nodes or organs. 7. Lymphatic dissemination does not necessarily relate directly to the site of the lesion, the depth of invasion, or the histologic grade. but it ma\ also depend on other probably complex factors. Admitting an inadequate understanding of the complex svstem. we have long used the euphemistic term "biologic predeterminism" to account for the manv as vet tlllcxpiained activities. Nodes Identification of the lymph nodes is more difficult and less complete than in the l!Suallv younger cervical cancer case. Failure to identify entire groups of nodes was not infre4uent. Involvernent of the rnajor lyrnph
node groups in this series, listed according to frequencv (Fig. 2). arc the aortics. hypogastrics. external iliacs, common iliacs, obturators. sacrals. mediastinals. inguinals. and the supraclavicular nodes.
Volume 123 :-lumber 1;
In 15 per cent of the series, only the aortics. with or without distant metastases, were involved. There was no remarkable difference between the treated and the untreated cases (Tables III and IV). Observations include the following: I . Identification of the nodes in the endometrial carcinoma case , frequentl y affected by surgical procedures, medical diseases and aging process, is not only difficult but at times incomplete. 2. Lodgement of a tumor embolism, within either a lymph vessel or a lymph node, may cause a partial or complete blockage which will interfere with the normal lymph flow. This in turn may shunt the lymph stream into alternate channels. These alternate channels probably represent previously rarely used routes rather than new channels. Though an occasional pattern suggests the possibility of retrograde flow, it is more likely that the collateral vessels permit dissemination to extensive and often bizarre sites. 3. The absence of an occasional node group in .. volvement. situated along a main channel. suggests that a small tumor embolism can pass either through or around a node, or even bypass the entire group. 4. A dehnite decrease occurs in both the size and the readily recognized number of nodes with increasing age: however, careful dissection occasionally demonstrates very small atrophied nodes. This would suggest there is not necessarily a decrease in the actual number of nodes. 5. Both microscopic and grossly recognizable changes in the lymph nodes are frequent. Such d escriptive terms as fatty infiltratio n , fibrosis, hyalinizatio n , devitalized ghost-cells, and many others have been e mployed. Such changes, as expected, are more common in this series than in the cervical cancer cases. T h ere was little difference between the treated and the untreated cases in this regard . There was no consistent alteration attributable to the effects of radiation . 6. The smallest positive node was 2 mm. in size and was accidentally included in a large tissue block. The largest node measured 2.5 em . and was located in the deep inguinal group. It was benign and probably a residual reaction to a previous inflammatory lesion of the vulva. 7. The mediastinal nodes can be free of disease in the presence of pulmonary involvement. and vice versa. 8. Widespread involveme nt of the pelvic nodes ca n occur without positive aortic nodes or distant metastases. and vice versa. 9. The appearance of distant active metastases 7, 12 , and 15 years following surgery for Stage I, Grade 1 disease must assume an entrapment of a malignant tumor embolism. If this occurs. the question arises as to
Lymphatic dissemination in endometrial carcinoma
573
Fig. I. Lymph vessels and lymph nodes of the cervix and body of the uterus. (From Henriksen. E.: AM . .J. 0BSTET. GYNECOL. 58: 924, 1949.)
how long such an embolus of malignant cells <.:an remain potentially viable a nd what is the cause of the resurgence. It is difficult to correlate some of the microscopic patterns, in the positive nodes. with their viability. The variations may reflect the method of tissue fixation, blocking a nd cutting techniques, along with marked variations in the methods of staining. Pyometra
Pyometra was diagnosed in 20 per cent of the series. It is likely that some cases were overloo ked . Actually , a small uterine cavity , containing 5 c. c. or less of purulent exudate, can be as potentially dangerous as a larger a nd therefore readily recognized case. In several cases it was thought in retrospect that radium had been i~s erte d in the presence of a small undiagnosed pyometra, and the patients died of septicemia. There is no evidence linking pyometra with the origin of cancer, but the associated vessel and tissue reactions might accelerate the spread of the malignancy. Distant metastases
Distant metastases were prese nt in 62 per cent of the cases. Contrary to preconceived ideas. there was little difference between the treated and the untreated cases. The many sites emphasize the importance of hematogenous spread; howe ver, the sites of the distant
574
.\H\.l 'lllhcr 1.-,, I ~J/ .-l A111. J. Ohstet. Cnwnll.
Henriksen
!NFRf{lUENT SITES
BRAIN OR SKULL. 5%
POST. NeCK NOIJ!S
HANIJ FEMUR
T/8/A
SUPRACLAVICULAR 12%
PIJ6/C 80/'lc
~c/JIAST!NAl
WNC
18 %
29%
PUIJ!?A
12%
AXIUAElf
S%
UVER 29% NOIJES
AI?R!NAL /4%
AORTICS 59%
KII?Nctf 10 % SPlEEN 14%
AORT!CS 0/'ltf/ 15%
PfRITONEIJ/of 39%
H{/POvA5TRICS 61%
IJR£TER Bro VERTEBRA 20% 80WCI. 29% OYARlf 34%
EXT. IL lACS 48% COM.ILIACS 40%
08TURATORS 37% SAC!?Al
22%
!Nt:iU!NAlS 16%
8tA/J11EE 23 %
VAviNA
/4%
VAt{. Ct/FF II% Vt/lVA 4°/o
Fig. 2. Relati ve incidence of distant metastases and lymph node involve ment in 188 cases ot endometrial carcinoma.
metastases are reported with no explanation offered for their itinerary. h must be stressed that the data are not complete, since a necropsy including all the possible sites is both impractical and impossible. If the mediastinal nodes and the pleura are included. the lungs represent the most com mon site. Other organs, listed in order of their frequency (Fig. 2). arc the liver, adrenal gland, kidney. and spleen. The ovary was involved in 34 per cen t of the cases. The vertebra was the most common site for bon y metastases with such sites as the hand, skull, pubic bone, pelvic bone. femur, and tibia noted infrequent!}'·
Previous surgery and malignancies Previous pelvic or abdo minal surgery is more frequent than in a similar number of cervical cancer cases. The resulting adhesions and scar tissue distortions may play an important part in interfering with the usual lymph How. These changes are also affected by previous medical diseases and the increased age of the average case with endometrial carcinoma. The appendix, often in association with other operative procedures. was removed in 83 (44 pe1· cent) cases. Various suspension proced ures occurred in 39 (21 per cent) cases. In 6 (3 per cent) cases both rubes and ovaries
were removed without th e ute rus. The gallbladder was removed in i7 (41 per cent) cases. T he exact epidemiologic importance of this high incidence is not clear but could reflect the unstable medical status of many of the cases with endometrial carcinoma. Previous surgery and /or radiation was done for ca ncer of the breast (4 per ce nt). colon (4 per cent). bladder (2 per ce nt), and stomach (2 per cent). !\inc granulosa-theca cell tumors were observed in 225 of the necropsies.
The assigned causes of death Ninety-two (49 per cent) cases died from causes not directly related to the endometrial carcinoma. Cardiovascu lar diseases and diabetes were the leading medical ca uses. Twenty-seven of the 92 cases died of other primary maligna ncies. In 96 (51 per cent) cases death was attributed to the e ndom etrial carcinoma. In order of freq uency the assigned causes were carcinomatosis which included distant me tastases. postradiation complications. he morrhage . kidney complications. and immedia te radiation com plications . The author is indebted to Dr. Em anuel A. Friedman for his editorial assistance.
Volume 12'\ Numher fi
Discussion DR. ARTHUR T. HERTIG, Southboro, Massachusetts. It is an honor and pleasure to discuss this amazing paper on a subject which has interested my friend and colleague and me, lo these many years. His initial paper over 25 years ago before this Society was on the same general theme. He is one of the masters of the subject. Some of his conclusions are, indeed. startling and unexpected. I believe that we have all grown up with the general concept that if one must have or treat a female genital cancer, that of the endometrium is a nice one; it seems not to be so. The 188 patients, all of whom were autopsied (52 by the author) and the remaining 136 randomly selected by him from over 700 cases, were arranged by Stages I to IV, histologically graded l to 4, and separated into those treated and untreated. My first comment is: How randomly were these cases selected? Was it on the basis of personal clinical contact with the patients? Was it because of some particularly interesting facet? Was it only those patients on whom complete autopsies were performed? Was a random series of numbers assigned to the 650 plus. nonpersonal autopsied cases and the 136 then chosen? Much depends upon the answer to this question--one I have already asked the essayist some days ago. At first I was amazed by Table I that there seemed to be no significant difference in the sites of metastases, whether the patients were treated or untreated. And, moreover, that cancer deaths were of essentially equal occurrence (50 per cent) in the two groups. Does this reflect that many of the treated cases were essentially untreatable when first seen? The essayist in his written paper indicates that in retrospect the treatment may have been insufficient or unsatisfactory. (Some of the cases go back many years.) Since Table I indicates the little difference between treated and untreated patients. it seems reasonable (although startling) to find in Table II that all of these patients, staged and graded, treated and untreated. were combined. Taking just one item, "distant metastases," it seemed to make very little difference what the grading and staging were as to whether distant sites were involved. To be sure, in Grade l, the spread, about 30 per cent, between Stages I and IV, respectively 49 vs. 89 per cent, is significant. As might be expected, the more extensive the lesion the more distant metastases there should be and, indeed, were found. Nevertheless, the amazing 49 per cent of Stage I, Grade I tumors. treated and untreated, with distant metastases. is both astounding and surprising. This is the key to the intrinsic treachery of the disease. At this point one is tempted to ask whether any of the Grade I, Stage I cases had the endometrial cancer confined to the endometrium? True cancers can be restricted to the endometrium. whether or not they would ultimately have invaded the myometrium. In Table III these figures prove to be 62 per cent for treated and ·!2
Lymphatic dissemination in endometrial carcinoma
575
per cent for untreated patients. This group of 37 patients is about 20 per cent in the entire series. Table IV shows, with respect to distant metastases. that histologic grading is of some significance as is the surgical or autopsy staging of the uterine cancer. Nevertheless, the fact that comparable figures for treated and untreated cases are so similar is disquietmg. It is to be noted that the peritoneum and ovary are likely and prominent sites of metastasis. One wonders if these two sites. with their rich vascular and lymphatic supply, are the answer to the distant metastases which leave no intervening footprints? Is the spread by the lymphatic duct from peritoneal implants valid? The concept of zones in the uterus which may explain bizarre spread is fascinating and needs further study. Could the tongues of malignant tissue invading the myometrium be analogous to the benign invasion of basal endometrium leading to adenomyosis? Dr. Henriksen and I have corresponded extensively over this paper. He has indicated that it may be more philosophic than factual. I doubt it. He is to be complimented for this laborious and painstaking study, which removes endometrial cancer from the "nice" to the "vicious" variety. Further study, probably along immunologic lines, will indicate why this cancer can do so many unexpected, and at present. anatomically illogical things. DR. CLYDE L. RANDALL, Buffalo. New York. I would like to comment on two points, for this is one type of gynecologic malignancy with which I have had sufficient experience to have gained personal convictions. Over a 30 year period we followed two groups wherein the results of somewhat over 100 patients treated one way were compared with an equal number treated by a different protocol. The first 200 were an attempt to evaluate the preoperative use of irradiation; the second series was an attempt to evaluate a more extensive operative procedure, with or without preoperative irradiation. The only conclusion I could reach was that in both studies too many patients were in older age groups in which deaths from causes unrelated to their adenocarcinoma of the uterus could obscure differences that might have appeared significant, had no patients in either group died of causes unrelated to the malignancy. The second point is that Carl Javert published a paper a number of years ago in which data were presented to demonstrate that endometrial carcinoma should be expected to exhibit a slowly progressive course. during which the treatment that was employed, whether early or relatively late in the life cycle of that particular neoplasm, really did not significantly alter the rate or the manner by which the neoplasm eventually accounted for the patient's death. ]avert's concept was particularly disturbing to Dr. Lewis Scheffey, who had emphasized for years that treatment would be
576 Henriksen
'()\l'lllbl'l
\Ill . .J. ( >h:.,tcl.
significanily more erreuive II adequarely planned and carried out according to plan. I am sure had J avert been here he would have expressed his appreciation of the studies Dr. Henriksen has reported this morning. DR. ERLE HENRIKSEN (Closing). In explaining my random selection of the cases I would like to inte1ject an old adage which holds that ''if the traveler has no idea where he is going, am road wiii take him there.·· There was no organized or so-called scientific selection, since complete necropsy n1atcrial on endometrial carcinoma is difficult to find. Through the cooperative efforts of many colleagues, the scrounging for protocols and slides, and a persistent alertness for possible material, the reported data were eventuallv collected. The selection of each case was based on the availability of all of the pertinent information, the thoroughness of the protocol preparation. and, in many instances. familiarity with the technical efficiency and interest of the in\'olved pathologist. The incidence of cases with the disease limited to the endometrium was unexpectedly low. On the other hand, in private practice, two patients without demonstrable invasion died of distant metastases. Admittedly. despite careful study. microinvasions could have been missed. I believe but cannot prove, hmn:·ver, that dissernination via the lyn1phatils or the blood vessel~ can originate in the endometrium. The relativek high incidence nf
1.1. 1~17.) (;\Ill'<
ol.
of peritoneal involvement is too low. On opening the abdominal cavity there is a tendency to overlook the peritoneal surface unless gross disease is present. The idea of the corpus being divided into poorlv defined tones or compartments is based on impressions gained from the examination of large free-cut sections. Dr. Hertig's suggestion that the extension of the tongues of the invading disease is analogous to the extension of aderwmyosis is logical. HowenT. it can be obserYed in ihe sante large section that sonte of the extensions suggest an orderly growth "·hereas in other areas the picture is disorderly and consistent with actual invasion and destruction of the mvometrium. This is consistent with the presence of mTrlapping compartments. I would like to re-emphasize the difficulties and pitfails encountered in this type of statistical stuciv. There are no definite ground rules and dinical staging ren1ains an enigmatic gamble. I an1 retnindcd of the "six blind men and the elephant. .. Though each described one part of the animal. none described the entire elephant. For man v years, endometrial carcinoma was nmsidered a "good cancer" and unfortunately many still hold to this idaea. It was also held that metastases were rare exccpi In the late stages of ihc disea~e. 'fhis. despite a report before this group in 1949, describing aortic node involvement and distant meta~tases present in the earlv stages of the disease. The preconceived idea that the priman site of the disease, the depth of invasions, and the histologic grade are dependable guidelines is not supported bv this report. However, the data do emphasize the potential ,·iciousness of the disease.
Ninth Annual 1-'ostgraduate ~ourse: Gynecoiogic Pathoiogy, Cytogenetics, and Endocrinology The Ninth Annual Postgraduate Course in Gynecologic Pathology, Cytogenetics, and Endocrinology will be held at the Sheraton-Mayfair Motor Inn, Milwaukee, Wisconsin, from March 31-April 7, 1976. This six-day cou;·se. which is limited to 150 applicants, is designed primarily as a postgraduate refresher course for residents, practitioners, and specialists desiring a current review of the pathophysiology of the reproductive tract. A $400.00 enrollment fee, payable to The Medical College of Wisconsin, will include 60 selected Kodachrome 35 mm. slides. The registration fee is nonrefundable. Guest faculty wiii include H. W. Jones, Jr., M.D., Georgeanna S. Jones, M.D., E. R. :\Tovak, M.D., J. D. Woodruff, ~LD., H. J. Davis, M.D .. and C. G. Julian, M.D. For further details and registration, \\Tite to: Richard F. ~y1atting1y, ~~,LD., T'hc ~"1cdical College of \\lisconsin, 8700 W. Wisconsin Ave., Milwaukee, Wisconsin 53226.