CERTAIN CYTOPLASMIC INCLUSIONS OF LIVER CELLS* ALWIN M. PAPPENHEIMER, M.D., AMS JoHx J. HAwTHoRmE, M.D. (Fromthe Department ofPathology, College ofPhysicians andSurgeons, Columbia University, Ne-x IYork City) On routine examination of liver sections our interest has been aroused by the rather frequent occurrence of certain, well character- ized cvtoplasmic inclusions. Although easily seen, even with ordi- nary stains, they seem to have attracted little attention and to have escaped the recognition of experienced pathologists. They are not described in standard textbooks ofpathology, histology or cytology. Mallory,1 however, in his classical paper on necroses of the liver (I90I) has accuratelv described and pictured these bodies (Plate XVI, Fig. i. and Plate XVII, Fig. i), and since we have found but one other reference to them his description may be cited in full. "In a majority of the early cases the protoplasm of the liver cells containsnumeroussmallandlargevacuoles inwhich are single small hvaline globules; single, sometimes multiple, coarse, and fine threads; and occasional networks, all of which stain deeply with eosin and with Weigert's fibrin stain. With other stains thev react exactly as fibrin does. The hyaline globules are not threads cut across; whether they really are fibrin is difficult to prove. They often seem to precede the formation of the threads and networks of fibrin." It would seem that MIallory regarded these bodies as an accom- paniment of early degeneration, ending in necrosis of the liver cells. Wehave been able tolocate butone other short reference to these structures. Taniguchi 2 (I93i) describes in liver cells three types of cytoplasmic indusions as follows: Type i: Spherical bodies, varying in size from a round mitochon- drial granule to that of a nucleus, homogeneous and with sharp contours. They stain deeply with Heidenhain's iron hematoxylin after fixation in Regaud's fluid and can be distinguished, though less clearly, in hematoxylin-eosin preparations. They do not stain with Sudan III. Type 2: More orless elongated oval bodies which are regarded as a modification of Type I. They are infrequently found. * Received forpublication March 25, 1936. 625 626 PAPPENHEIMER AND HAWTHORNE Type 3: Comparatively narrow, long and short, thread-like, slightly bent, sometimes spindle shaped bodies. They are derived from swollen, thread-like mitochondria. The material for Taniguchi's study was obtained from surgical biopsies. The bodies described as Type i were found in 27 of 94 cases. The photographic illustrations accompanying the paper show, though not very clearly, spherules and filaments. We have been interested in making a somewhat more extended study of these bodies. DESCRIPTION OF THE INCLUSIONS (A) Spherical Bodies: Our observations in the main confirm the description given by Mallory. The spherules vary in size from that of a smallmicrococcusup to 4or 5 ,uin diameter, themajority rang- ingfrom I to 3 ,u. Theylie in a large vacuole which, withthe usual staining methods, is optically empty or contains only a little shred- ded material. When the vacuole is contiguous to a nucleus the nuclearmembranemaybeindentedbyit. Theremaybebutasingle spherule or the entire cytoplasmmay be riddled with vacuoles, each containing a single body, so that the cell has a cribriformorporous structure. Adjacent vacuoles may coalesce into large spaces con- taining several spherical inclusions. When they are on the surface they may rupture, discharging the inclusion body into the space between liver cell and sinus wall. Staining Reactions With hematoxylin-eosin staining, the spherical bodies are acido- philic, but the color is unlike that of the cytoplasm or of the ery- throcytes, having a brownish yellow cast. They are sharply con- toured, moderately refractive, and entirely homogeneous without any suggestion of internal structure. WiththeGramstainthemethylvioletorgentianvioletisretained, ifdecolorization isnotpushed toofar. Thedegreeof Grampositive- ness is about the same as that of the nucleolus. Mallory's phosphotungstic acid hematoxylin stains the spherules very selectively a bluish black. After formalin fixation, staining with phosphotungstic and hema- toxylin demonstrates that the spherule is surrounded by a homo- CYTOPLASMIC INCLUSIONS OF LIVER CELLS 627 geneous capsular material which stains a dull brick red and com- pletely fills the vacuole (Fig. 3). Heidenhain's iron hematoxylin also brings the bodies out sharply but they are resistant to decolorization with ferric chloride. With azocarmine, anilin blue and orange G, the bodies take a dull orange color similar to that of the erythrocytes. With methyl green-pyronin, both spherules and rods take only a faint yellowish stain. With mucicarmine the globules do not take the mucin stain. Masson's trichrome stain shows the spherules to be red, in con- trast to the dark violet gray of the cytoplasm. The nucleoli are bluish black. Contents of the vesicle are pale blue. There is no un- stained vacuole. The bluish material is sometimes homogeneous, sometimes finely granular (Fig. 4). The rods stain orange-red and lie in a dear vacuole without the bluish staining matrix. Mallory's anilin blue collagen stain shows about thesamepicture as that of Masson's, except that the spherules are orange-red and the bluish matrix is less sharply brought out. With Mann's stain the spherules are dull red to bluish purple, depending on the degree of differentiation. Filaments stain about the same. Contents of vacuoles remain unstained (Fig. i). With Giemsa, they are stained a rose pink. Themost selective stain that we have found is Laidla-uw's method for demonstrating virus inclusions.* By careful differentiation with orange Galcohol the cytoplasm of theliver cellsmay bedecolorized to a grayish blue: the spherical inclusions retain the fuchsin and stain an intense crimson. The red blood cells are a tawny orange- red (Fig. 2). NVhile Laidlaw's method calls for fixation in acetic * LAIDLAW'S METHODFOR ST.AIING INCLUsION BODIES irF.ixation in gslaatcuarlataecdetaiqcueacoiuds c.5orrcocs.ivesublimate ............... IOOcc. 2. Embedinparaffinandcutat 3 . 3. Deparaffinize in xylol, absolute and 9 7% alcohol andrinse inwater. 4. Stainin Weigert's ironhematoxylin (2%) 5 minutes. 5. Differentiate ino.,7c acidalcohol. 6. Rinsein tapwaterfollowedbydistilled. 7. Stain in I%c aqueous acidfuchsin 5-I5 minutes. 8. Rinseindistilledwater. 9. Mordant in i7c phosphomolybdic acdfor30 seconds. io. Rinsein distilled water. ii. Differentiatein0.25%c ornge Gin 70% alcohol. 12. Dehydrate, clearand mountin balsam. 628 PAPPENHEIMER AND HAWTHORNE acid bichloride, Zenker's fixation without acetic acid gives equally brilliant pictures. The bodies are not stained by scharlach R. (B) Rod Shaped and Filamentous Structures: These occur almost always in association with the spherical bodies, but far less fre- quently (only in about one-fifthof the cases). They also lie in clear vacuoles and assume a variety of forms-simple bacillus-like rods, beaded or fuzzy straight rods, or long curving fibrils, sometimes looped upon themselves. Small crossing or interlacing filaments may be seen within the same vacuole. Not infrequently thefibrils seem to be formed directly on orfrom the spherical body; the spherules may lie in the middle or at the end of the rod. But in many of the vacuoles containing rods or filaments, no spherical body can be discerned. The staining reaction of these filaments and rods, as Mallory pointed out, is the same as that of fibrin. They retain the gentian violet in the Gram-Weigert stain, and stain bluish black with phos- photungstic acid hematoxylin. They are also brought out by the Heidenhain iron hematoxylin method and by the Laidlaw stain. General Incidence In all, 562 sections of liver from human cases have been studied. The spherical inclusions were found in 175 cases, or 31.I per cent. Many blocks were recut and stained with phosphotungstic acid hematoxylin, and in these preparations the bodies are more easily recognized. Their incidence in the hematoxylin-eosin sections was therefore somewhat lower (23.7 per cent) than in the I24 sections stained with phosphotungstic acid hematoxylin, which gave 43.6 per cent of positives. Includedinthe 562 casesare39I adults, I24children, and47 new- born infants and stillbirths. The percentage incidence in the differ- ent age groups is shown in Table I. TABLEI wks. 2wks.- I-5 6-io II-20 21-30 31-40 41-50 5I-60 61-70 7I-80 80+ Age 2WS. Iyr. yrs. yrs. yrs. yrs. yrs. yrs. yrs. yrs. yrs. yrs. No. 47 46 30 i8 30 42 66 79 104 65 32 3 No.+ 14 6 7 6 9 14 27 23 40 17 IO 2 %+ 29.7 I3.0 23.3 33-3 30.0 33.3 4I.0 29.1I 38.4 26.I 31.2 66.6 CYTOPLASMIC INCLUSIONS OF LIVER CELLS 629 There is, as the analysis shows, no significant difference in the various age groups. The spherules were found once in a stillbom full'term infant, and in several prematures. Occurrence in the Human Fetus Through the courtesy of Dr. Coler of the Sloane Hospital, we havebeen abletoobtainmaterial fromhumanfetusesfrom5months to term. The sections stained withphosphotungstic and hematoxy- lin showed an astonishingly high incidence of inclusions. AgeofFens No.Ezaimied Positive Negative or Dou.5tfil 5mos. 7 7 6mos. 7 4 3 7mos. 7 6 I 8mos. 6 , I Ful term 7 4 3 34 26 8 The failure to identify the bodies in a few of the cases may have been due to poor preservation of the tissue. In 8 of the 34 cases, that is in about the same proportion as in the postnatal livers, the spherules were associated with rods and filaments. Relation to Possible Postmortem Changes The bodies seem to be relatively little affected by autolytic changes. and they do not becomemore numerousafter death. They have been found as early as 45 minutes portmortem and as late as 53 hours. The incidence of positive findings at various periods after death is shown in Table II. TABLE II Hourspostmortem 2 .2-6 6-12 12-24 Morethan 24 Totalautopsies .............. 26 1i65 1I30 I46 43 No. showingglobules ......... 6 4 39 i6 No. showingrods ............ I 12 I3 4 9 %C globules + ............... 23.I 30.3 34.6 26.7 37-2 %rods+ ................... 3.8 7.2 I0 6.2 9.3 Although there are slight variations in the different groups, there is no trend toward adefinite increase ordecrease in the incidence, as the period elapsing after death becomes greater. 630 PAPPENHEIMER AND HAWTH0ORNE Relation to Other Diseases We have attempted to correlate the occurrence of the spherules with various general diseases such as arteriosclerosis, nephritis, and acute and chronic infections. As might have been anticipated from theratheruniform age distribution andthepresenceof thebodiesin the fetal liver, no such correlation seems to exist. The bodies have been present in all sorts of pathological conditions, and a detailed analysisofourmaterialfromthispointofviewwouldbefutile. The only reservation to the above statement is in the case of malignant disease. Amongthe 562 casesexamined, iii hadmalignant tumors, and of these 24 showed liver inclusions- an incidence of 21.6 per cent as against an incidence of 33 per cent in the 45I non-tumor controls. Thesignificanceofthisloweredoccurrenceisnotapparent, butthedifferencewouldappeartobetoo greatto ascribe to chance. Not only is there no clear-cut correlation with general diseases, but we have been able to find no constant association with any particular type of liver lesion. The inclusions are found frequently in otherwise normal appearing cells. When they are very numerous thecytoplasmofthecellscontainingthemhasaspongyorcribriform appearance. They are alwaysmore abundant in the central portion of the lobule and often seem to be especially numerous when the liver cellsare atrophic, or inthepresenceof central congestion. But they are quite definitely not associated with necrosis of the cells and we cannot interpret their presence as an indication of early degeneration or cell death. On the other hand, they may often be recognized in cells that have recently undergone necrotic changes. We have notfound them in the neoplastic cellsof primary liver cell carcinoma. Occurrence in Laboratory Animals In monkeys the spherical inclusions have been found in I of 5 livers examined. In guinea pigs they have been seen in 8 out of 44livers- slightlylessfrequently than in human livers. In 7 fer- ret livers they were present in every instance. Rods and fila- ments, however, have not been observed. On the other hand, we have searched in vain for these bodies in the liversof rabbits, dogs, cats, pigs, rats, mice, chickens or ducks. CYTOPLASMIC INCLUSIONS OF LINER CELLS 63I DIscussIoN It has not been possible to arrive at any final conclusion as to the nature of the spherical inclusions. Several possible interpreta- tions come to mind and may be discussed briefly: i. They are some unknown structural component of the cell, like the Golgi net, centrosome, or mitochondria. 2. They are secretory products. 3. They are degenerative products, akin to the colloid droplets in degenerative renal epithelial cells. 4. They are cytoplasmic "virus inclusions." It is dlifficult to believe that these structures represent some un- known organs of the cell. Were such the case, one would expect to find them in every cell, certainlv in everv liver and in every species. Nor is there any reason to believe them to be derived from known cv-toplasmic constituents. They stain quite differently from mito- chondria, persist long after the mitochondria have been destroyed by autolvsis, and are not affected by acid. There is nothing to sug- gest that they are related to the Golgi net or centrosome, andthey are most certainly not due to extrusion of nuclear chromatin. We can find no particular evidence for or against the possibility that they represent a secretory product of the liver cells. The fact that with Gram and phosphotungstic acid hematoxylin the staining is like that of fibrin. brings up the thought that they may be asso- ciated with the production of fibrinogen. But the stainingforfibrin is far from being specific and it would require much more evidence to justify such an assumption. Against the secretory nature of the product is the observation that only a single body may be present in the cell. We can recall no instance inwhich secretory granules were notmultiple. Perhaps it may be taken asa point also against the secretory nature of these bodies that they are not found in all human livers, as one might expect if they represent a normal activity of the liver cells, and that they are not demonstrable in the livers of dogs, rats, rabbits, cats, chickens or ducks. Against the viewthatthespherules representadegenerative alter- ation in the cvtoplasm is the fact that they are often found in cells that shownootherevidenceofinjury. It is true that they aremost numerous in the central portions of the liver lobules, where atrophy 632 PAPPENHEIMER AND HAWTHORNE or necrosis accompanying central congestion is so frequently seen. Doubtless also, their presence in great number, with the resulting spongy rarefaction of the cytoplasm, may lead to the distintegra- tion of some of the cells. On the other hand, the bodies may be foundin cellsadjacent totheportal spaces, and in thissituation the liver cells often show no evidence whatever of damage. The last possibility which comes to mind is that the bodies are of the nature of virus inclusions. This assumption would lead of necessity to the view that alargeproportion of human livers harbor a non-pathogenic virus or viruses, and that similar viruses occur in the livers of monkeys, guinea pigs and ferrets. The fact that we have found the inclusions in the liver cells of the fetus would also logically involve the assumption that the hypothetical virus is transmitted to the embryo. The only evidence in favor of these bodies being virus inclusions is (i) their morphology, which is not unlike that of certain cyto- plasmic indusions, such as the Guarnieri bodies (cf. Cowdry, E. V., J. Exper. Med., 1922, 36, 666, Fig. 33); and (2) the staining reac- tions. We have hadopportunity to compare theliver cell inclusions with cytoplasmic virus indusions in the various tissues of ferrets experimentally infected with a virus now being studied by Dochez, Slanetz and Smetana.* The staining of the liver cell inclusions is practically identical with that of the virus inclusions, but the latter are not surrounded by a stainable halo. It would be of interest to obtain experimental support for the virus theory, as the demonstration of such a frequently occurring, latent, non-pathogenic virus would be of great theoretical impor- tance. Unfortunately wehave thoughtof noobviouswayofattack- ing the problem experimentally at the present time. As regards the rod-like and filamentous inclusions, we agree with Mallory that they are probably fibrinous in nature. They occur in about one-fifth of the livers containing the spherical inclusions, but only in association with them. They may occupy the same vacuole as the spherical bodies, and indeed one frequently gets the impres- sionthatfilamentshaveformedonthespherical bodiesasanucleus. But often the rods and filaments occur in vacuoles that contain no spherical body. The only plausible explanation for their occur- rence is that coagulable fluid from the plasma has seeped into the * Weareindebted toDr. Smetanaforplacingthisferretmaterialatourdisposal. CYTOPLASMIC INCLUSIONS OF LIVER CELLS 633 vacuoles from the adjoining sinusoids and has crystallized out within the vacuoles. Why this does not occur in every case is not easy to understand. SUIBARY AN-D CONCLUSIONS A study has beenmade of certain spherical, rod-likeand filament- ous cytoplasmic inclusions that occur frequently in the liver cells of man, monkeys, ferrets and guineapigs. Theyhave notbeen seen intheliversofotherlaboratory animals. Whilenodecisionhasbeen reached as to their nature and significance, various possible inter- pretations are discussed. REFERENCES i. Mallory, F. B. Necroses of theliver. J. M. Research, igoi, 6, 264-279. 2. Taniguchi, A. Cytological studies of the liver cells. Tr. Jap. Palh. Soc., I93I, 21, 260. DESCRIPTION OF PLATES PLATE II7 FIG. i. Livercellscontainingsphericalandfilamentouscytoplasmicinclusions. Mann's stain. FIG. 2. Spherical inclusions. Laidlaw's stain. FIG. 3. Spherical indusions surrounded by a homogeneous, slightly reddish substance. Formalin fixation, phosphotungstic acid hematoxylin stain. FIG. 4. Spherical inclusions surrounded bybluish staining material. Masson's trichrome stain.