Vol. 9, No. 1, pp. 12–23

© 2002 Lippincott Williams & Wilkins, Inc., Philadelphia

“Similar confessions could be made about other pathologic,

or for that matter, clinical entities, for which we

are all now on the alert.”

Dr. Novak was not a formal pathologist, as we understand

it today. He practiced clinical gynecology in Baltimore

and because he had helped Dr. Cullen, who had

organized the gynecologic pathology laboratory at Johns

Hopkins, he was now its director—a common situation

in the early years of surgical pathology. It is therefore not

surprising that pathologists of his time viewed him with

reservations and that his gynecology peers did not let

him admit patients to Johns Hopkins (4).

Looking back, his sincere, humble, and wise comment,

as well as the magnitude and quality of his contributions

to science, leave no doubt that he knew very

well the trade of pathology. He deserves the recognition

that is owed to him by our community of pathologists.

Since my first report, the original description has been

widely confirmed and mentioned in the literature (references

in [5]), although errors and inaccuracies occur in

some textbooks and publications. Common misconceptions

are to consider the change, exclusively, a hypersecretory

phenomenon, or a nonspecific regressive alteration

due to fetal death, trophoblastic disturbance, or

decreased hormonal levels (6,7). Some have, equivocally,

stated that “the occurrence of mitosis in carcinoma

and its absence in the Arias-Stella reaction is important

for the differential diagnosis” (8). This article summarizes

the most recent contributions to the literature, both

mine and those of other investigators, and attempts to

clarify the pathogenesis.

The main characteristic of the reaction is cellular enlargement,

mainly of the nucleus, to double or many

times normal size. Without the presence of nuclear enlargement,

the phenomenon cannot be diagnosed. Hypertrophied

nuclei can show an ovoid or round shape with

granular or vesicular viable chromatin, an irregular outline

and a hyperchromatic appearance, or a compact,

pyknotic pattern (Fig. 1).

Mazur et al. (9) have used the term “optically clear

nuclei” to describe an aspect of the alteration, which is

characterized by centronuclear vacuolization resulting

from the replacement of the chromatin by a net of fine

filaments. Another morphologic variant is the occurrence

of intranuclear cytoplasmic invaginations—pseudoinclusions.

One or another of these alterations of the

macronuclei has been confused with herpetic endometritis

(10) (Fig. 2).

Usually the alterations are focal; they involve a group

of glands or only some part of them. Occasionally, they

can be quite extensive or “florid.” It is important to emphasize

that the presence of hypersecretory glands alone

does not constitute the phenomenon. The presence of

hypersecretory foci in the gestational endometrium was

described by Opitz (11) at the beginning of the last century.

Even though I originally described two forms of the

alteration (2) and later recognized a third form (12), accumulated

experience allows us to distinguish the five

histologic variants: 1) minimal atypia; 2) early secretory

pattern; 3) secretory or hypersecretory pattern; 4) regenerative,

proliferative, or nonsecretory pattern; and 5)

monstrous cell pattern. These variants are a function of

comparison with the phases of normal endometrium and

degree of atypicality.

Minimal atypica is the pattern usually seen at the beginning

of gestation. The nuclear enlargement is minimal

and occurs in limited foci (Fig. 3). Because the decidual

reaction is absent or is not conspicuous in the early stages

of gestation, the diagnosis of minimal atypia may have a

special practical meaning. This is what we have found in

some women who have been evaluated for infertility

(Table 1).

The early secretory pattern resembles normal early

secretory endometrium at day 17–18 of the menstrual

cycle. In these cases, nuclear enlargement is marked and

the endometrial cells show subnuclear or subnuclear and

supranuclear vacuoles (Fig. 4). The nuclei are centrally

>

located. The cells are frequently arranged in a palisading

fashion, with the nuclei at the same level. The affected

glands can show intraluminal papillary projections.

Since the original description, my colleagues and I

pointed out the occurrence of mitosis in the phenomenon

and, more recently, that they can be normal and abnormal

(Tables 2 and 3) (13). It is precisely in this model of

presentation that they are more frequent and abnormal

(Fig. 5). It is understandable why this histologic variant

is the one most likely to be confused with adenocarcinoma.

In my experience this model has been common in

cases of uterine abortion and ectopic pregnancy.

Secretory or hypersecretory pattern is the classically

recognized form of reaction. The glandular cells display

intense and diffuse cytoplasmic vacuolization, which

predominates as a distinct morphologic feature. The enlarged

and hyperchromatic nuclei are usually pyknotic

(Fig. 6). Focally, one can distinguish cells that are less

vacuolated and show a dense cytoplasm (Fig. 7), which

is due to their different histochemical composition.

The comparative epithelial immunohistochemical profile

of the normal functional endometrium with the atypical

change has demonstrated that the epithelial membrane

antigen (EMA) and CK7 produce an intense and

homogenous reaction in both normal endometrium and

the Arias-Stella reaction (ASR) (Table 4). The vacuolization

that characterizes the hypersecretory pattern of

atypical change highlights the increased membranous localization

of staining with these markers, creating a

“chicken-wire” pattern (Fig. 8).

Ultrastructural studies done by de Brux (14), Thasher

and Richart (15), and Salazar and Burgess (16) have

shown few organelles and sparse particles that resemble

granules of ribonucleic acid in the clear cells, and a conspicuous

Golgi apparatus, vesicles of endoplasmic reticulum,

numerous mitochondrial crests, and Palade

grains in the dark cells. One detail that has been found

repeatedly is the presence of parallel rows of rough endoplasmic

reticulum (Fig. 9), which is a characteristic

described in various tumors of Mullerian origin. This

variant is found mainly in uterine abortions.

In the regenerative, proliferative, or nonsecretory pattern

there is usually no evidence of secretory activity, or

it is minimal in the glandular cells. The enlarged nuclei

display a vesicular configuration or a granular chromatin

with a well-delineated nuclear membrane. The glands are

similar to those observed in the proliferative or regenerative

endometrium (Fig. 10). I have found this variant

in cases of hydatidiform mole, choriocarcinoma, ectopic

pregnancy, and uterine abortion.

In rare cases, the monstrous cell pattern, which is usually

focal, affects the whole endometrium. The histologic

section is dominated by the presence of giant and bizarre

nuclei, which involve all the cells in the glands (Fig. 11).

The atypical nuclei show a dense, homogeneous chromatin

and, frequently, pseudoinclusions. This model of

atypia originates problems of histologic interpretation. In

our experience, my colleagues and I have found it in

0.5% of florid alterations.

Even though the alteration was originally described in

the functional endometrium, which is more sensitive to

hormonal effects, today we have proved that when stimulation

is intense, the change can also occur in less sensitive

basal regions. From the beginning my colleagues

and I pointed out that not only the glandular but also the

covering epithelium could be affected and that the alteration

occurred in both the endocervix and the tubal epithelium

(17,18).

In recent years our observations and those of many

authors have demonstrated that under the stimulation of

chorionic tissue—gestational condition or trophoblastic

proliferation—the change can be observed in many territories

and lesions, including endometriosis (peritoneum,

subcutaneous, umbilical) (19); endocervical polyps

(20,21); vaginal adenosis (22,23); germinal inclusion

cysts (ovary) (24); para-ovarian and para-tubaric cysts

(25); and mucinous cystoadenoma 26).

It is interesting that Albukerk and Berlin (27) have

described nuclear changes that resemble the Arias-Stella

reaction in the luteal cyst of gestation and that Clement

and Scully (28) described similar atypias in luteinized

follicular cysts of gestation and puerperium.

Two questions that must be answered are: How early

in the gestation and how late after delivery does the

endometrial atypia occur? Fortunately, studies by

Holmes and Lyle (29) and Dahlerup and Jorgenson (30)

and the review by Oertel (31) of material that Hertig

examined to elucidate the early stages of human embryogenesis

have answered these questions.

The accumulated information shows that in certain

clinical situations, finding the Arias-Stella reaction can

be especially significant, as an isolated histologic sign, or

as part of a group of changes, in making a diagnosis.

According to our experience and that of other investigators,

the change has been of histologic diagnostic significance

for recognizing early uterine pregnancy (12), ectopic

pregnancy (32,33), trophoblastic tumors (indirect

evidence), and postabortion or postpartum metropathies.

Histologic differential diagnosis can be problematic in

the following conditions: metastatic adenocarcinoma

versus foci of endometriosis in pregnancy (5); endometrial

adenocarcinoma versus florid Arias-Stella reaction

(8,34); adenocarcinoma versus Arias-Stella reaction in

fallopian tube (35); and atypical mucinous cystoadenomas,

germinal inclusion cysts, and paratubal cysts in

pregnancy and puerperium (26).

It is interesting to note that in recent years there has

been special emphasis in finding the change in the endocervix

and recognizing atypical cells in Papanicolaou

smears.

Schneider (36) has found the alteration in the endocervix

in 17 of 191 uteri, which were removed during

gestation, proving that the proximal segment of the cervical

canal was the most frequent location. Cove (37) and

Rhatigan (38) have described cases of endocervical

Arias-Stella reaction that caused confusion with premalignant

or malignant conditions. The same problem was

encountered by Cariani and Guderian (39) and McCormick

and Menaci-Williams (40) in endocervical polyps

with Arias-Stella reaction. McCormick and Menaci-

Williams reported the occurrence of tripolar mitoses,

confirming our finding of abnormal mitosis in the reaction.

In my experience the endocervical Arias-Stella reaction,

whether in normal mucosa or polyps, frequently

shows the monstrous cell pattern (Fig. 12) and gives rise

to the differential diagnosis with neoplastic processes.

The gestational state is often ignored and it is the debate

over the nature of the cervical lesion that reveals the

pregnancy condition.

Hilrich and Hipke (41) were the first to show that

atypical endometrial cells could appear in Papanicolaou

smears. Since then, many observers have verified the

same finding (42,43). Albukerk and Gnecco (44) pointed

out that the abnormal cells in the Papanicolaou smear

could be confused with neoplastic cytology and that they

are found more frequently in cases of ectopic pregnancy,

in which the lack of a uterine-placental barrier makes it

easier for desquamation and passage of the cells to the

vaginal fundus.

In recent years, Pisharodi and Jovanoska (45), Mulvany

et al. (46), Yates et al., (47), and Michael and

Esfahani (48), have reported observations showing that

the abnormal cytology due to Arias-Stella reaction, of the

endometrium or the endocervix, in pregnant women,

originates problems of cytologic interpretation. We have

verified this occurrence in and Figure 13 illustrates one

such case.

Recently, because of similar experiences, Benoit and

Kiny (49) recommended including Arias-Stella reaction

cells among the atypical glandular cells to be considered

in Papanicolaou smears from women in pregnancy or

puerperium.

An understanding of the mechanism of enlargement of

the nucleus and the unique histologic reaction is essential.

Three basic facts must first be mentioned:

1. Nuclear hypertrophy is due to an increase in the contents

of DNA. The finding of double Barr corpuscle

by Dahlerup and Jorgensen (30), the karyometric

studies of Chiara (50) and the Fulgen microspectrocytophotometric

determinations of Wagner and Richart

(51) early established that nuclei enlarge

because of polyploidism and that aneuploidism does

not exist.

2. The aspect of the histologic reaction is evidence of

occurrence of proliferative and secretory stimuli, acting

simultaneously.

3. The change can be present in normal physiologic conditions,

including term pregnancy (52), postpartum

(30), therapeutic abortion (51,53), and early normal

pregnancy (12).

The glandular changes in the endometrium are a consequence

of the action of estrogens and progesterone.

The former stimulate cellular proliferation and the latter

stimulate secretory activity (54,55).

With this background, since my first observations I

postulated that estrogen and progesterone are, in some

way, involved in the pathogenesis of the change. Early

on, using human chorionic hormones and estrogens, I

was able to obtain, experimentally, proliferative and secretory

activity simultaneously in normal rats, and, using

estrogens and progesterones in castrated rats to induce

the same changes. In these experiments I also observed

focal nuclear enlargement (56,57). In 1966, experimental

studies by Dr. Dallenbach (58) confirmed these findings.

Since then I have performed some experiments on

humans. Administering a combination of estrogens and

progesterone to postmenopausal women who were going

to have hysterectomies, I was able to induce, after 1

month of treatment in some of them, endometrial

changes that were focal and suggestive of the Arias-

Stella reaction (5).

In another experiment, due to the action of high doses

of estradiol and hydroxyprogesterone, I was able to produce

focal changes that were similar to the endometrial

atypia (5).

Conversely, since the observations of Dockerty et al.

(59) and Azzopardi and Zayid (60), it has been recognized

that therapy with synthetic progesterone-estrogen

can induce endometrial changes of the Arias-Stella reaction

type. Recently, Huettner and Gersell (61) have reported

the induction of what they consider Arias-Stella

reaction in eight postmenopausal or perimenopausal

women who received treatment with progestational hormones.

Since the classic studies by Good and Moyer on the

doses of estrogens and progesterone, it is possible to

obtain, experimentally, secretory or proliferative endometrial

responses (62).

All of the above supports, my proposition that the

pathogenesis of the phenomenon depends on a balance of

these two hormonal effects.

How do we continue exploring along these lines of

thought?

Today, we know more about the hormonal mechanisms.

In the case of female hormones, we know that

there are cellular receptors, that estrogens stimulate the

synthesis of DNA, RNA, and proteins, by a chain reaction

mechanism, and that progesterone reduces the synthesis

of DNA and RNA (63–65). We know the cell

cycle and we can identify, in tissues, some of the factors

of intermediation (Fig. 14). Therefore, there is room for

further investigation. We have investigated, using immunohistochemical

methods, the presence of estrogen and

progesterone receptors and markers of cellular proliferation,

Ki-67 and proliferating cell nuclear antigen

(PCNA), in 20 cases of florid Arias-Stella reaction, compared

with findings in the normal phases of the endometrium

and in endometrial hyperplasia.

In the absence of an accepted standardized method for

evaluating immunohistochemical reactions, we have

adopted the method recently proposed by Allred et al.

(66) for scoring the immunostaining signal for estrogen

receptors and progesterone receptors in breast cancer. In

each case the best stained fields were chosen for the

evaluation. In the Arias-Stella reaction we found that the

estrogen receptors and the progesterone receptors (Table

5) had a total score less than in the normal functional

phases and in hyperplasia.

Scores for the cellular proliferation markers (Table 6)

were also less than in the proliferative phase, but nevertheless,

higher than those of the normal secretory phase.

To summarize, the results of the immunohistochemical

study show that:

1. Estrogen and progesterone receptors (Fig. 15) are

present in the foci of the endometrial atypia.

2. The positive reaction for the proliferative factors,

such as antigen Ki-67 and PCNA (Fig. 16), is also

present in these foci. It should be stressed that estrogen

receptors, progesterone receptors, Ki67, and

PCNA have been found present in enlarged typical

Arias-Stella reaction cells (Fig. 17).

These results are, in essence, in agreement with the

immunohistochemical study recently reported by Doss et

al. (67).

Steroid hormones can act by either of two ways: the

nongenomic mechanism or the classic, or genomic, form

(68). The nongenomic effects are fast—occurring in seconds

or minutes—are highly specific, do not require

nuclear receptors, and do not induce RNA or protein

synthesis.

The nature of the Arias-Stella reaction corresponds to

that of a long duration process—days, weeks, or

months—and as demonstrated above, it is accompanied

by the involvement of receptors and growth factors that

lead to DNA synthesis. Therefore, the evidence supports

the view that steroid action in the Arias-Stella reaction

follows, basically, the classic genomic path.

All of these findings suggest that the pathogenesis is

related to the effects of estrogen and progesterone acting

simultaneously, but as highlighted in the immunohistochemical

study, with a higher estrogen weighting than

found in the normal secretory phase. This imbalance

would be a determining factor, through intermediary

mechanisms, to induce the increased synthesis of DNA

that leads to the occurrence of polyploidism.

It is interesting to note that recently, the late Mexican

Professor Márquez-Monter et al. (69) compared the macronuclei

of the Arias-Stella reaction to those seen in

megaloblastic anemia, suggesting that in some cases,

they result from a change in the cellular cycle by mutation

of the “check point” bound to the mitogenic factor,

which determines the continuation of DNA synthesis. It

cannot be ruled out if in these hormonal interactions

there is a role for factors directly derived from chorionic

tissue.

How do we locate this alteration from the perspective

of cell biology? The characteristics that we have highlighted

reveal an adaptive, controllable, and reversible

cellular reaction induced by hormonal actions that comprise

proliferative and antagonistic stimuli. Because the

cells that modify are mature cells that are not replaced, it

is not a form of metaplasia.

If the variability in the shape and size of the cells, the

pleomorphism and hyperchromasia, and the presence of

atypical mitosis could suggest dysplasia, the DNA polyploid

contents, absence of aneuploidy, and absence of

progression would not correspond with this process (Fig.

18, Table 15) (70).

At this level of knowledge, the accumulated information

leads us to think of a singular form of transdifferentiation:

conversion of a differentiated cell into another

differentiated cell. In this case it would be called “atypical

transdifferentiation” (Fig. 18).

This atypical transdifferentiation, which is characterized

by polyploidism, absence of aneuploidism, and absence

of progression of lesions, and which is found in

tissues that are sensitive to antagonistic hormonal factors,

would also occur in other areas, including atypical

cells in seminal vesicles (71,72); monstrous cells in epididymis

and duct deferens (73,74); atypical cells in lutein

cysts in pregnancy (27); atypical cells in luteinized

follicle cyst of pregnancy and puerperium (28); and bizarre,

benign cells in thyroid (dyshormogenetic goiter)

and other endocrine organs (75).

All of the above information allows us to conclude

that the description of endometrial atypia, associated

with the effect of the chorionic tissue, has had until now,

practical significance in gynecologic and obstetric pathology.

Owing to its peculiar relation to the synthesis of

DNA, it is possible that elucidating the pathogenesis of

the changes will have even greater importance in the

scope of cell biology. This is the challenge for future

generations.

1. Arias-Stella J. Historia de un Epónimo Médico. Cuento para Estudiantes

de Medicina. To be published.

2. Arias-Stella J. Atypical endometrial changes associated with the

presence of chorionic tissue. Arch Pathol 1954;58:112–8.

3. Novak E. Surv Obstet Gynecol, p. 440 1–2, comments to

reference (1), 1955.

3. Novak E. Surv Obstet Gynecol, p. 440 1–2, comments to

reference (1), 1955.

reference (1), 1955.

4. Carter D. Surgical pathology at Johns Hopkins. In: Rosai J, ed.