Journal of Clinical Laboratory Analysis 5293-298 (1991)

Tissue Peroxidase in the Normal and Neoplastic Salivary Gland Donald Armstrong,' Dale Van Wormer,* and Sandra Dimmitt3 Department of Medical Technology, School of Health Related Professions, SUNY at Buffalo, Buffalo, New York;' Tulsa Medical College, University of Oklahoma,' and Department of Pathology, Hillcrest Medical Center, Tulsa, Oklahoma Peroxidase activity was measured by two colorimetric methods in six neoplastic tumors of the parotid salivary gland. When p-phenylenediamine was used as the hydrogen donor co-substrate,peroxidaseactivity was absent or reduced from 6 to 22% of normal in these tumors. Cross reactivity with normal peroxidase antibody could not be demonstrated in tumors exhibiting negative activity. Two neoplastic salivary glands had well-described, encapsulated tumors which showed reduced peroxidase activity when compared with the normal uninvolved portion. However, when guaiacol was substitutedfor p-phenylenediKey words:

adenoma, immunochemistry, parotidgland, saliva, preoperativetesting

INTRODUCTION Peroxidase is present in the saliva of most mammals (1) including man (2,3) and is mainly glandular in origin (4). It is a main enzymic constituent of these glands and accounts for 1-2% of the total protein in some species (5). The enzyme is composed of several isoenzymes (6,7), and is immunochemically identical to lactoperoxidase (8). Synthesis of peroxidase occurs within acinar cells, in the perinuclear cisternae, in endoplasmic reticulum, and on ribosomes where it is secreted into the salivary duct along with the secretory granules (9). The release of endogenous salivary peroxidase is mediated via cyclic AMP (lo), or by isoproterenol (1 I). Salivary peroxidase also appears to be organ specific (12- 14) as the number of isoenzymes (6,15), sub-cellular localization (9,16), stability (17), and sensitivity to inhibitors (1,9,18) differ greatly from peroxidase obtained from other tissues. Physiologically, glandular peroxidase is known to act by catalyzing the iodination of tyrosine (19-21). In saliva, it has an antimicrobial effect combining with thiocyanate and peroxide to inhibit growth in the mouth (22). A relationship between low peroxidase activity and neoplasia has been reported in myelogenous leukemia where myeloperoxidase is deficient (23,24). The incidence seems to be higher in males, and other oxidative enzymes such as catalase and d-amino acid oxidase are normal (25). Peroxidase cytochemistry has been used to selectively identify different lines of acute myelogenous leukemia (26). As a result 0 1991 Wiley-Liss, Inc.

amine, normal activity was recorded. After dialysis, peroxidase activity was not enhanced. Mixing experiments showed no effect of the tumor extract on the peroxidase activity of normal saliva, nor on salivary gland supernatant. Normal salivary gland peroxidase was inhibited 90% by lO-'g/L aminotriazole, but neoplastic tumors were inhibited less (13 and 73%). It is suggested that perioxidase is a simple marker for the detectionof neoplasia and can be of value in the differentiationof benign tumors.

of the enzyme deficiency, peroxides and their mutagenic by-products are increased in various types of cancer (27-29). In order to determine whether peroxides produced by tumors are the direct result of a reduction of a specific enzyme activity, we studied peroxidase and catalase in human salivary tumors. We chose these types of tumors because I ) peroxidase is very high in normal salivary glands and 2) because tumors of these glands are well circumscribed so that normal and neoplastic tissues could be studied simultaneously.

MATERIALS AND METHODS Salivary glands were obtained either surgically or at autopsy. Two cases (3 and 5) had well-encapsulated tumors and both normal and tumor tissue from the same patient were compared. All glands were cleaned of blood, connective tissue, and fat. The tissues were then frozen immediately and stored at - 15°C until the day of assay. After thawing, the glandularappearing tissue was dissected from the remainder of the organ, blotted between two layers of filter paper, and weighed. The samples were homogenized in a Dual1 glass homogenizer (4 volumes/gram) in 0.15 M phosphate buffer, pH 7.4. Once the specimens were uniformly dispersed, they were centri~

Received March 26, 1991; accepted April 2, 1991. Address reprint requests to Donald Armstrong, Ph.D.. Dr.Sc., Professor and Chairman, Department of Medical Technology, School of Health Related Professions, UB Clinical Center, AA 107.462 Grider St.. Buffalo, NY 14215

294

Armstrong et al.

For the immunological method, human parotid saliva (10 fuged at 100,000 rpm for 15 minutes, and the supernatant was decanted and used directly for the measurement of per- mL from each gland) was collected from a normal donor by oxidase activity. The incubation mixture consisted of 5- 10 using a Curby cup after stimulation with sour lemon candy. microliters of enzyme, 0.1 milliliter of 0.28 g/L p-phenyl- The parotid saliva was immediately concentrated to 50 X in a enediamine, 0.1 milliliters of 0.03 g/L, hydrogen peroxide and Minicon B-15 (Amicon Corporation) filter at 4°C and then 0.15 g/L phosphate buffer, pH 7.4 to a final volume of 3.2 frozen and stored at a - 70°C. Antisera to lactoperoxidase was obtained from New ZeamL (30,3I ) . The reaction was performed at 25°C and was initiated by the addition of hydrogen peroxide. The stable purple land rabbits immunized with purified bovine lactoperoxidase chromogen was recorded at 485 nm. A standard curve using (Sigma Chemical Co.) in complete freund adjuvant as folhorseradish peroxidase (Sigma Chemical Company, type 1 1) lows: 89 pg on day 1,66 p,g on day 2, and 66 pg on day 10. was used to determine units of enzyme activity. Specific activ- On day 28, the rabbits were reinjected intravenously with 90 ity was obtained by taking the optical density produced by pg of enzyme in 0.08 ml of the physiologic phosphate buffer each specimen following 100 seconds of incubation, convert- and bled 10 days later. Human parotid salivary peroxidase ing that value into nanograms of horseradish peroxidase, and was then immunochemically purified by precipitation with then dividing by total protein. This method of expressing cross-reaction bovine lactoperoxidase antibody. Template enzyme activity was chosen because two substrates are micro-double diffusion was performed on a thin layer (1 mm) required for the reaction to proceed (18). It was considered of 1% agarose on glass slides pre-coated with 0.2% agarose. more convenient, therefore, to express salivary peroxidase Following incubation at 4°C in a humid atmosphere, the preactivity relative to horseradish peroxidase activity, rather than cipitates formed were presumed to contain peroxidase. To conto calculate activity on the basis of hydrogen peroxide hydrol- firm the presence of peroxidase in the precipitate, the agarose ysis, p-phenylenediamine conversion, or p-phenylenediimine gel was washed free of unprecipitated protein with a washing formation. Peroxidase was also measured in some cases with buffer, blotted with filter paper to remove buffer salts, and then dried at 37°C according to the method of Johansson and guaiacol as co-substrate. Dialysis experiments were performed in running cold Hjerten (34). Uriel's reagent (35) was applied, rinsed away ( 8°C) tap water and were allowed to continue for 24 hours when the purple color appeared, and then dried once more. before assay. In some experiments, samples were dialyzed Tissues were also embedded in paraffin, sectioned, and stained in a closed system so that the contents outside the bag could with H & E for histopathological identification. Salivary gland tumors were typed and classified as recombe recovered. In other experiments, aminotriazole at a final concentration of g/L was added directly to each super- mended by the WHO International Reference Centre, Lonnatant before assay and compared to a duplicate sample with- don (36). out the inhibitor, Mixing experiments were performed by adding an equal volume of normal saliva to an aliquot from RESULTS the appropriate tissue. In a separate experiment, an aliquot of parotid gland from the normal portion of a tumor was Enzyme Analysis mixed with an aliquot from the well-encapsulated portion of the same gland. Six salivary glands, obtained from patients with parotid Acid phosphatase was measured by using p-nitro phenyl- tumors and from one patient with metastatic disease but withphosphate as substrate (Sigma Tech. Bull., No. 104) cata- out salivary involvement, were compared to normal parotid lase by the method of Luck (32) and protein by the method tissue (Table 1). Two pleomorphic adenomas (cases 1 and 2) of Lowry et al. (33). Hemoglobin, horseradish peroxidasewere completely devoid of peroxidase activity. Two other type 1, microperoxidase, and lactoperoxidase standards were pleomorphic adenomas (cases 3 and 4) were very low in purchased from Sigma Chem. Co., St. Louis, MO. peroxidase activity. The first two cases were more cellular

+

TABLE 1. Tissue Peroxidase and Control Enzymes in Parotid Salivary Gland Tumors Case No.

Diagnosis

Age & sex

I. 2. 3. 4. 5. 6. 7.

Normal (N = 7) Pleomorphic adenoma Pleomorphic adenoma Pleornorphic adenoma Pleomorphic adenoma Chronic sialoadenitis with inflammatory cyst Adenolymphoma Metastatic carcinoma from pancreas (autopsy)

Adult 41 M 55 M 46 M 31 M 66 M 67 M 81 F

"With p-phenylenediamine as co-substrate.

Peroxidase"

Catalase

Acid phosphatase

3,079(1,316-4,276) 0 0 I70 290 596 667

57 (25-80) QNS

23.1 (18.6-46.2) QNS

43 12.3

13.3 13. I QNS 26.9

1,408

-

58 -

-

Peroxidase in Neoplasia

295

TABLE 3. The Inhibition of Peroxidase Activity by Aminotriazole Source

B Inhibition

Standards Hemoglobin Horseradish-peroxidase Myeloperoxidase" Microperoxidase Amniotic fluid peroxidaseb Lactoperoxidase Tissues' Normal submaxillary gland peroxidase Normal parotid gland peroxidase Case 3 Case 5 Case 6

0 0 0

0 0 96.5 90.0 89.2 13.4

73.0 31.9

"Prepared as previously described (3I). bObtained by amniocentesis and prepared as previously described (32) 'Prepared from autopsy tissue as described in the text.

Fig. 1. Ouchterlony plate showing lines of identity between monospecific antiserum to peroxidase from normal parotid fluid (center well). a parotid extract (right well), and fresh parotid salivary juice (lower well). The top well contained an extract from the tumorous portion of case 3

When aminotriazole was added to the parotid extracts of patients 3, 5 , and 6, inhibition was only 41% (range 13-73).

with a greater epithelial component than the later two cases. The tumors from cases 5 and 6 were also well below normal in total activity, but higher than the pleomorphic adenomas. Finally, one patient with a primary carcinoma of the pancreas and liver (case 7) with extensive metastases to regional lymph nodes had normal salivary peroxidase activity. Catalase was low in only one case. In cases 3 and 4, peroxidase activity was normal in the unaffected portion of the gland, but markedly reduced in the tumorous portion. When guaiacol was used as co-substrate in place of p-phenylenediamine, there were no major differences between normal and neoplastic tissues (Table 2).

The activity of normal salivary peroxidase was not affected by dialysis (Table 4). After extensive dialysis, peroxidase activity remained unchanged in samples of cases 1 and 3. A small amount of activity was lost during dialysis in cases 4 and 6. Essentially no inhibition of the normal activity was observed when tissue extracts from the tumors of cases 1 and 3 were mixed with either an aliquot of normal saliva or an equal aliquot from the normal portion of the same gland. When the hornogenate from the neoplastic portion of case 3 was dialyzed in a closed system and the fluid outside the bag was collected, concentrated, and added back to the normal portion, there was no inhibition of the normal activity. Peroxidase was detected immunochemically in the normal portion of parotid gland tissues from cases 3 and 4 by using monospecific antiserum to parotid fluid peroxidases. In both instances, the enzyme was undetectable in the tumorous portion of the same gland. When cases 3 and 4 were concentrated four and five times respectively, and then re-immunophoresed,

Effect of Inhibitors Lactoperoxidase and normal salivary and submaxillary peroxidase activity were markedly inhibited by 0.01 g/L aminotriazole-i.e., 96.5,90.0, and 89.2% respectively (Table 3). TABLE 2. Comparison of Peroxidase Co-Substrates in Normal and Neoplastic Tissue Case No. Normal controls 3

4

Diagnosis and sample

p-Phenylenediamine Guaiacol peroxidase peroxidase"

Effect of Dialysis and Mixing

TABLE 4. Effects of Dialysis and Mixing on Parotid Salivary Gland Peroxidase Activity Case No. Treatment

Pleomorphic adenoma Normal portion Tumor portion Pleomorphic adenoma Normal portion Tumor portion

"Same conditions as in Table 1 .

3,079 ( I ,3 16-4,276)

.I2

Normal controls

2.648

.08

1

I70

.21 3

3,091 290

.II .I3

4

6

Dialysis Dialysis Mixed with normal pooled saliva Dialysis Mixed with normal portion of same gland Dialysis Dialysis

Change from original activity (%)

+ +

0 0 2.2

0 1.4

- 5.2 - 26.2

296

Armstrong et al.

vary gland supernatants and the dialysis experiment described earlier for case 3 confirms that peroxidase activity is virtually absent. Another explanation for decreased activity could be that as the tissues sustain increasingly more damage, membranes become ‘‘leaky’’ and peroxidase is lost in this manner. Oral, pharyngeal, and laryngeal carcinomas have recently been shown to leak Ig A into the saliva through damaged epithelium (39). When guaiacol was substituted in place of p-phenylenediDISCUSSION amine, the deficiency was not seen. This could mean that I ) Peroxidase was completely absent in the parotid gland of there are two peroxidases, each preferentially active against two cases of pleomorphic adenoma, even at 20 times the nor- one or the other co-substrate or that 2) some of the isoenmal assay volume. The peroxidase values of two other cases zymes known to be present in saliva (6)have a greater affinwere greatly reduced in activity and varied from 6 to 22% of ity for guaiacol. Therefore, isoenzymes with a rapid turnover normal. At the same time, catalase and acid phosphatase were rate against one substrate might hydrolyze as much, or more, within normal limits. In the tissues, much of the peroxidase as all of those present against a different substrate. The twoactivity was apparently not due to a true peroxidase, because and-one-half-fold increase in guaiacol peroxidase shown by aminotriazole did not appreciably inhibit the enzyme as it the tumorous portion of case 3 is suggestive of such a hypothgenerally does with normal salivary peroxidase. Thus, of esis. By comparison, this same tissue had a decrease of perthe 667 units recorded for peroxidase in case 6, only 253 oxidase activity (from 2,649 to 170) when measured with units were due to true peroxidase. The remainder might be p-phenylenediamine as co-substrate. While we were not able to measure peroxide levels due to due to the pseudo-peroxidase activity of various hemecontaining proteins (21). Of the 170 units recorded for case the small amount of tissue given us at the time of surgery, it 3, only 23 units were due to peroxidase; in case 4,212 out of is interesting to note that elevated levels of hydrogen perox290 units and in case 5,435 out of 596 units were due to true ide (ninefold) have been measured indirectly in leukemic (40) peroxidase. However, not all of the non-enzymatic activity cells. Those authors proposed that it was indeed due to the can be attributed to hemoglobin alone since case 1 (with zero decrease in peroxidase. Aromatic hydrocarbons, especially activity) had hemolytic contamination similar to the other five 3,4-benzopyrene, are carcinogenic (41), and this effect is marksamples. It would appear therefore that residual enzyme in edly enhanced by the addition of hydrogen peroxide in Fenton’s neoplastic tissue is capable of some degree of activity, as mea- reagent (H202 FeC12). That important study showed that sured against p-phenylenediamine or guaiacol, but has some- hydrogen peroxide rapidly decomposed to hydroxyl radicals how been structurally altered so that it does not recognize when Fe2+ is added, and the production of these radicals the antibody produced from samples that were inhibited dif- was considerable accelerated. The free radicals so generated ferently. Differences in net activity might be explained because could then react with benzopyrene to form oxy- or peroxywe are dealing with neoplasms of variable duration. In the 3,4-benzopyrene radicals. The former radical is known to be case of catalase-deficient tumors, the decrease is progressive converted into a naturally occurring metabolite, 5-oxy benwith the growth of the tumor (32). Thus, it is logical to assume zopyrene (42). Thus, if peroxidase were deficient, high perthat whatever mechanism is operable in reducing peroxidase oxides with or without iron present and some carcinogen, activity would do so to a greater extent in tumors of long could result in tumor formation which would be compatible standing. This information was not available to us in the pres- with the results of our present report. The incidence of ent study, but this point will hopefully be clarified in future autofluorescent lipofuscin granules are reportedly increased experiments. in neoplastic lesions of the salivary gland (58% of cases) A heat-stable, non-dialyzable inhibitor has been demon- whereas the number containing similar granules averages 45% strated in cells from leukemic mice (37). This does not seem in normal tissue (43). Since peroxidized lipids are considto be the case, however, in our present studies, because dial- ered one important component of lipofuscin, decreased perysis did not increase activity and normal activity was not inhib- oxidase activity could be expected to result in deposition of ited by addition of tumor extract. The severe loss of peroxidase more granules. activity in parotid gland tumors is all the more striking when Immunochemically, peroxidase from the normal parotid compared to nonnal proliferating tissues, where peroxidase gland forms a line of identity with parotid salivary peroxiis found in high concentrations (38). dase using a monospecific antibody. In comparison, tumor The peroxidase activity determined in case 7 where metas- tissue extracts from the same gland contained no immunotases were widespread would argue against circulating inhib- logically detectable peroxidase activity. itors produced in the liver, pancreas, or other tissues. The Only one other enzyme study has been reported in human inability of tumor extracts to inhibit saliva or normal sali- neoplastic salivary gland (44). It was found that increased

the line of identity was still missing. However, when the normal portion of salivary gland from cases 3 and 4 was diluted 1:100 so that total enzyme levels measured colormetrically were similar to the cancerous portion, a slight precipitation line was still observed. The presence of the enzyme in the precipitate line was confirmed by running a colorimetric reaction directly on the double diffusion plate.

+

Peroxidase in Neoplasia

amounts of alkaline phosphatase was present and associated with the nuclear envelope of epithelial cells. In conclusion, it is proposed from our present experiments that only one enzyme, peroxidase, or an isoenzyme with a high Vmax for p-phenylenediamine, is deficient and therefore is a useful marker for the detection of salivary neoplasia. Ultrastructural peroxidase studies (38) and cell surface markers (45) using specific lectin staining patterns can also be helpful. A combination of the immunologic and enzymatic tests applied to saliva as described in the study should provide a direct assessment for preoperative evaluation of patients with parotid gland tumors.

ACKNOWLEDGMENTS This work was supported by a grant from the Milheim Foundation for Cancer Research and by the Hillcrest Medical Center Education Fund.

REFERENCES 1 . Carlsoo B, Kumlien A , Bloom G: A comparative histochemical study of peroxidase activity in the submandibular glands of five mammalian

species including man. Histochemie 26:80-88, 1971. 2. Mosimann W, Sumner J: Salivary peroxidase. Arch Biochem 33:487, 1951. 3. Hojo Y: Selenium and glutathione peroxide in human saliva and other body fluids. Sci Total Environ 65%-94, 1987. 4. Nickerson J , Kraus F, Peny W: Peroxidase and catalase in saliva. Proc SOL.Exp Biol (NY) 95:405-409, 1957. 5 . Morrison M, Bright J, Jayasinghe W: Lactoperoxidase V. Identification and isolation of lactoperoxidase from salivary glands. Arch Biochem I II:126-133, 1965. 6. Iwamoto Y, Nakamura R, Tsunemitsu A, Matsumura T: The heterogeneity of human salivary peroxidase. Arch OrulBiol13:1015-1058, 1968. I. Mansson-Raherntulla 8, Rahemtulla F, Baldone D, Pritt K , Hjupe A: Purification and characterizationof human salivary pemxidase. Biochemisfry 27:233-239, 1988. 8. Morrison M, Steele W: Lactoperoxidase, the peroxidase in the salivary gland. In Biology of rhe Mouth. P Person, ed. American Association for the Advancement of Science, Washington, D.C., 1968, p 89-1 10. 9. Strum J. Karnovsky M: Ultrastructural localization of peroxidase in submaxillary acinar sells. ./ Ultrastruer Res 3 1:323-336, 1970. 10. Carlsoo B, Danielsson A, Marklund S , Stigbrand T: Effects of 3’,5’-cyclic adenosine monophosphate, 5-hydroxytryptamine, noradrenaline and theophylline on the simultaneous release of peroxidase and amylase from the guinea pig submandibular gland. Acfu Physiol Scand 91:203-210, 1974. 11. Yamashina S, Barka T: Peroxidase activity in the developing rat submandibular gland. Lab lnvesf 31:82-89, 1974. 12. Rytomaa T, Teir H: Relationship between tissue eosinophils and peroxidase activity. Nature 192:271-272, 1961. 13. Steimeszynska T, Zgliczynski .I:Studies on hog intestinal mucosa peroxidase. E u r J Biochem 1956-63. 1971. 14. Haldar I, Mahajani V, Datta A: Comparative study of peroxidases from the thyroid and the submaxillary gland of goat (Capra hiscus). Comp Biochem Physiol44B:511-527, 1973. 15 Felberg N, Schultz J: Evidence that myeloperoxidase is composed of isoenzymes. Arch Biochem Biophys 148:407-413, 1972. 16. Hosoya T, Matsukawa S, Nagain Y: Localization of peroxidase and other microsomal enzymes in the thyroid gland. Biochemisty 10:3086-3092, 1971.

297

17. Alexander N, Corcoran B: The reversible dissociation of thyroid iodide peroxidase into apoenzyme and prosthetic group. J Biol Chem 237: 243-248, 1962. 18. Brill A, Weinryb I: Reactions of horseradish peroxidase with azide. Evidence for a methionine residue of mouse thyroid and submaxillary gland. Biochemistry 6:3528-3535, 1967. 19. Banerjee R, Chowdhury R, Datta A: Effect of erythroprotein on the peroxidase and tyrosine-iodinase activity of mouse thyroid and submaxillary gland. Eur J Biochem 34: 169- 176, 1973. 20. Hati R, Datta A: Studies on the enzymatic formation of di-iodotyrosine by a microsomal preparation of goat submaxillay gland. J Endocrinol 44:177-185. 1969. 21. Alexander N: Iodide peroxidase in rat thyroid and salivary glands and its inhibition by antithyroid compounds. J Biol Chem 234: 1530- 1533, 1959. 22. Iwamoto Y, Inoue M, Tsunemitsu A, Matsumura T: Some properties of the salivary antibacterial factor (S. A. factor). Arch Oral Biol 12: 1009- 1012, 1967. 23. Catovsky D, Galton D, Robinson J: Myeloperoxidase deficient neutrophils in acute myeloid leukemia. Scand J Haematol9: 142-148, 1972. 24. Uliyot J, Bainton D: Azurophil and specific granules of blood neutrophils in chronic myelogenous leukemia: an ultrastructural and cytochemical analysis. Blood 44:469-482, 1974. 2 5 . Hokama Y. Kimura L, Kohara T, Perreira S, Su D, Torikawa C , Oishi N: Variations in peroxisomal enzyme levels of peripheral leukocytes of cancer, leprosy, and tuberculosis patients. Cancer Res 34:2784-2789, 1974. 26. Sundstrom H, Korpeia H, Viinikka L, Kauppila A: Serum selenium and glutathione peroxidase, and plasma lipid peroxidases in uterine, ovarian or vulvar cancer, and their responses to antioxidants in patients with ovarian cancer. Cancer Left 24: 1-10, 1984. 27. Wyss 0, Wiessen C, Schaiberger G: Peroxidases and mutations. Radial Res[Suppl] 3:184-191, 1963. 28. Basu A, Mamett J: Unequivocal demonstration that malonaldehyde is a mutagen. Carcinogenesis4:331-333, 1983. 29. Shamberger R: Increase of peroxidation in carcinogenesis. J Narl Cancer Inst 48: 149I - 1497, 1972. 30. Armstrong D, Van Wormer D , Dimmitt S: Studies in Batten disease. Arch Neurof 30:144-152, 1974. 31. Armstrong D, Van Wormer D, Dimmitt S , May P, Gideon W: The determination of peroxidase in amniotic fluid. OB-GYN 47593-598, 1976. 32. Luck H (ed.): Cafalase. Academic Press, New York, 1965, p 885. 33. Lowry 0, Rosembrough N, Farr A, Randall R: Protein measurement with the Folin phenol reagent. J Biol Chem 193:265-275, 195I . 34. Johansson B, Hjerten S: Electrophoresis, crossed immunophoresis. and isoelectric focusing in agarose gels with electroendosmotic flow. Anal Biochem 59:200-213, 1974. 35. Uriel J: Characterization of enzymes in specific immunoprecipitates. Ann NYAcadSci 103:956-977, 1963. 36. Thackray A, Sobin L: Histological 7jping ofSalivary Gland Tumors. World Health Organization, Geneva, 1972. 37. Strauss R, Paul B, Selvaraj R, Sbarra A: A peroxidase inhibitor in leukemic AKR mouse spleen cells. Cancer Res 34:3220-3224. 1974. 38. Neufeld H, Levay A, Lucas F, Martin A, Stotz E: Peroxidase and cytochrome oxidase in rat tissues. JBiolChem 233:1958, 1958. 39. Brown A, Lally E, Frankel A: IgA and IgG content of the saliva and serum of oral cancer patients. Arch OralBiol20:395-398, 1975. 40. Malmquist J: Serum myeloperoxidase in leukaemia an polycythemia Vera. Scand JHaematol9:311-317. 1972. 41. Nagata C, Tagashira Y, Kodama M, loki Y, Oboshi S: Effect of hydrogen peroxide, Fenton’s Reagent, and iron ions on the carcinogenicity of 3,Cbenzopyrene. G A “ 64:277-285, 1973.

298

Armstrong et al.

42. Nagata C, inomata M, Kodama M, Tagashira Y: Electron spin resonance study of the interaction between the chemical carcinogens and h u e components. III. Determination of the structure of the free radical produced either by stirring 3,4-benzopyrene with albumin or incubating it with liver homogenates. GANN 59: 289-298, 1968. 43. Buchner A, David R: Lipofuscin in salivary glands in health and disease. OrulSurR 46:79-86, 1978.

44. Cutier L, Chaudhry A, Montes M: Alkaline phosphatase activity associated with the nuclear pore in normal and neoplastic salivary gland tissue. JHistochem Cytochem 22:) 113-1 117, 1974. 45. Daley T, Tolson N, Wysocki G: Lectin probes of glycoconjugates in human salivary gland neoplasms. J Oral Pathol 14:531-538, 1985.