Document VGmeN7XgBGZXE26dXrz3ezw8K

O c? C cr> JD Allan R. Batterman and Philj? M. Cook U S. Environmental Protection Agency. Environmental Research Laboratorv-Duiuth. 6201 Congdon Boulevard. Duluth. M.\ 55804. USA BaTTERMAN, A R.. AND P M Cook. 1981. Determination of mineral fiber concentrations m fish tissues. Can. 1. Fisn. Aquat. Sci 38: 952-959. Submtcroscopic inorganic particle concemrations in (issue have not been quantitatively determined in the past because of a lack of sample preparation techniques capable of achieving the sensitivity required. The determination of whether mineral fibres in w ater are accumulated in aquatic organisms requires transmission electron microscope examination of bulk tissue residues rather than thin sections The simple preparation method used for this investigation involved removal of water and organic matter by freeze-drying and low temperature ashing. Lake trout with a lifetime exposure to Lake Superior water containing amphibole fibers contained similar amphibole fibers particularly in the kidney and with low concentrations in muscle tissue. Lake trout from two locations with widely different water fiber concen trations had corresponding differences in tissue fiber concentrations. Analysis of other fish raised under laboratory condiuons suggests that ingestion is the primary route for fiber accumulation. Key words: electron microscopy, fiber accumulation, asbestos, lake trout, brook trout, channel catfish, .Arctic char Batterman, A. R., and P. M. Cook. 1981 Determination of mineral fiber concentrations in fish tissues Can. J. Fish Aquat. Sci. 38: 952-959. Le manque d'une technique de preparation des echantillons permeuant d'attcindre la sensibility rcquisc est la raison pour laquelle les concentrations dc particulcs inorganiques submicroscopiques dans les tissus n'ont pu etre mesurees quamitativement dans le pass Afin de determiner si des fibres mindrales presentes dans I'eau s'accumulcnt dans les orgamsmes aquatiques, il faut examiner au microscope electronique des tissus en vrac plutdt que de minces coupes Dans I'ytude ddente ci-dessous. la preparation des echantillons comporte I'cnlevemem dc i'eau et de la matiere organique par cryodessication et incineration a bassc temperature Des touladis ayant ete exposbs toute leur vie a I'eau du lac Supdneur contenant des fibres d'amphibole ont ygalcment de ccs fibres, particulidrement dans le rein et, a dc faiblcs concentrations, dans le tissu musculaire. Des touladis provenant de deux endroits dont la concentration des fibres dans I'eau ytait ends differcntes montrtrent des differences correspondanles de concentration des fibres dans les tissus. L'analyse d'autres poissons Sieves dans des conditions de laboratoire donne a penser que I'ingeslion est la pnncipale vote d'accumulation des fibres. U~i CD O CO Received December 4, 1980 Accepted April 24, 1981 Re?u le 4 ddeembre 1980 Accepte le 24 avnl 1981 Recent concern over the human carcinogenicity of asbestos and analogous mineral fibers when ingested (Lee 1974) has spurred interest in the development of methods for the quanti tative determination of their presence in tissues. The discov- ery of mineral fibers in human unne has demonstrated that some fibers can pass through the human gastrointestinal tract following ingestion of water contaminated with amphibole' fibers (Cook and Olson 1979). The same type of fibers were 1 Amphibole minerals are members of a class of hydraled silicates having a double-chain crystal structure These minerals can crys tallize dunng rock formation in an asbesnform habit tbundles of individual fibers) Amphibole minerals used commercially as asbes tos (when mrnable as the asbesnform variety I are amosite (fibrous grunemc), crocidolitc, anthophyliite, tremolite, and actinolue Most amphibole particles in western Lake Supenor water fall in the cum- Printed in Canada (J63I0) Impnmb au Canada (J6310) mmgtoniie-grunente senes and appear to range from nonasbestiform to asbesnform in ongin A fiber in a water sample is defined as any panicle with a iength-to-widih ratio equal lo or exceeding 3 I Same nonasbestiform amphibole mineral crystals, when crushed, rorm cleavage fragments that art microscopically very similar to. or idcnucal with, small fibers, which result from the crushing of amphibole asbestos, in morphology, crystal structure, and chemistry About 95% of the asbestos used m North America is chrysolite, a fibrous serpentine mineral with a distinct tubular microstructure 952 Species SalveUnus nama\cuih S. m2mj\cush S alpinus S. fonttrtahs taalurus iacustns batterman and cook mineral fibers in fish 953 Table I Fisn samples analyzed Location Split Rock. Minnesota Huron Bay. Michigan Deception Bay, Hudson Bay. Quebec ERL-D. Duluth. Minnesota ERL-D, Duluth. Minnesota Age (years) 3-4 4-5 7-12 Collection date imo/yr/ May 1978 March 1980 Sept 1978 2-3 5-6 Feb 1980 May 1976 Estimated avg waier fiber concr (fibenL) No of Anphiboie fish X 10* Chrysoule X J0* > 100 <1 i 1 Unknown Unknown 0-500 Food source Natural diet Natural diet Natural dicr i 50 50 < 1 Lab diet (Glencoe* PeLeis) < 1 Lab diet (Glencoe* Pellets) recently reported to be present in human liver, lung, and jejunum tissues as a consequence of ingesting Lake Superior water containing amphibole fibers (Carter and Taylor 1980). No studies of the accumulation of fibers by aquatic organisms have been reported with the exception of the optical micro scope observation of particles in mussels (MytUus edulis) exposed to asbestos tailings (Halsband 1974). Because min eral fibers contaminate some natural waters and fish arc thus exposed to these fibers, we have investigated the possible accumulation of these fibers in fish tissue. Optical microscope examination of tissue preparations for mineral fibers is not practical because most of these particles have widths less than the resolving power of the optical microscope (Berkeley et al. 19651. Electron microscopy of standard histological preparations of ashed thin sections is not feasible because the approximately 70-nm-thick tissue sections would require astronomical numbers of fish in tissue to see one fiber in an entire section (Cook 1979) Since detection of concentrations of only a few hundred fibers or less per gram of tissue is required, a method is necessary which separates the fibers from larger pieces of tissue without altering them This method must be able to eliminate all organic matter without changing the morphological, chemical, or physical character istics of fibers. To determine if freshwater fish used as human food can accumulate mineral fibers in their tissue, either by contact with the particles in contaminated waters or by ingestion of contaminated organisms of the food chain, several species of fish with different exposure histones were selected for analysis. Methods Exposure concentrations for fish collected from natural waters (Table I) are approximations based on the assumption that the fish spent a significant portion of the year prior to capture in the collection area rather than distant areas with radically different fiber concentrations All fish except the brook trout were frozen for storage anchor shipment prior to thawing and removal of tissue samples for analysis Brook trout were collected from laboratory stock at the time of tissue preparation. Arctic char samples were obtained from frozen tail sections received after other portions of the fish were analyzed for toxic chemicals (1 Boulva and A. LeBcau. Department of Fisheries and Oceans, C.P. 15500, Quebec, Que. GIK 7Y7, 1979. personal cummumcation). The fish were dissected in a filtered air laminar flow hood in a clean room with a positive-pressure particle-free air sup ply Dissecting instruments were cleaned with particle-free water One set of instruments was used to remove the skin, another set to remove a block of ussue. and a third set to remove a subsection of tissue for actual analysis. The tissue sample was placed in a 25-mL Pyrex sample tube, weighed, frozen in liquid nitrogen, freeze-dned for -- 5 h, and placed in a low temperature asher (LTA) at 150 W and 1.5 mL Oj/min for at least 16 h. The ash was suspended in ~ 15 mL of water acidified with 6 drops of 6 m HC1. Blank samples were prepared by processing an empty tube with the tissue sample tubes in each chamber of the LTA. The ash suspension was placed in an ultrasonic bath for 2 mm, mixed on a vortex stirrer for I min, and filtered throuah a 25-mm 0.1-p.m Nuclepore membrane filter. The filter w is placed in a plastic Petri dish and dried for -- 30 min a: j0C. One-half of the filter was carbon-coated in a vacuum evaporator. Small pieces (2 mmJ) were cut out of the carbon-coated filter and placed carbon side down on Formvar-film-coated electron micro scope finder grids. The grids were placed on a piece of coarse metal screen on top of several layers of filter paper in a Pctn dish. A small drop (10 p,L) of 0.1-p.m filtered chloroform was placed on each filter section and then the layers of filter paper were immediately saturated with filtered chloroform. The Petri dish was covered, and the chloroform allowed to dissolve all of the filter piece and Formvar film for -- 12 h. The grid which resulted was covered with a thin carbon film in which sample particles were imbedded (Fig 1) This tech nique prevents the loss of panicles during transfer of the sample from the filter to the grid, preserves the distribution of panicles originally in the sample, and leaves a replica of the filter (0.1-p.m holes) which can be used to check on folds or breaks in the carbon film which would create areas of uneven particle distribution Randomly chosen grid openings with areas of -- 9500 p.m! were systematically scanned at a magnification of 10 OOOx or higher on a JEOL 100C electron microscope in the trans- STOCK) 4 188 954 CAN J FISH AQL'AT SCI . VOL J8. IV8] ST00l.lt I 89.. e Fig I. Electron micrograph of a fiber, I 6 (im long, identified in a lake trout kidney preparation The selected area electron dtffractio; pattern Uowtr left nse-.) established the mineral group as amphibole The energy-dispersive X-'av specltum itonrr right inseni with majo silicon and iron peaks established the mineral as grvnente mission mode. Grid openings with broken carbon film or other disruption of the sample were rejected Each particle with an aspect (length to width) ratio si was identified by campanson of its morphology, selected area electron diffrac tion (SAED) pattern, and energy dispersive X-ray (EDX) spectrum with those obtained for standard mineral samples. Amphibole (Fig I) and chrysotile fibers produce character istic SAED patterns and EDX spectra when oriented properly in (he electron beam. Fiber size, orientation, or interfering particles cause difficulties with positive identification of fiber mineralogy. Fibers which possess amphibole or chrysotile morphology but SAED patterns and/or EDX spectra not clearly diagnostic for amphiboles. chrysotile. or some other mineral were classified as "ambiguous " Each sample was examined until 100 particles of interest were counted or at least 10 gnd openings were scanned. Fiber concentrations were calculated from the formula. No. of fibers/mg no. of fibers x filtration area (urn') area examined (p.mJ) x weight of sample (mg) Fiber size distributions were analyzed and compared by com puter program. Elemental ratios calculated from integrated EDX peak intensities were plotted on ternary diagrams to facilitate the comparison of mineral fiber elemental com positions for different samples. The tissue preparation technique dcscibed above was chosen after attempting several other techniques involving chemical digestion. The principle objective was to optimize the sensitivity of the analysis by increasing (he amount of tissue processed per unit area of electron microscope gnd examined. Digestion chemicals included Soluenc*. KOH. HjO;, and a mixture of K0H-H2O2 Although these reagents all effectively digested muscle, kidney, and liver tissues, the digests when filtered through a Nuclepore filter (cither 0 1-p.m or 0.4-p.m pore diameter) left too much residue to allow direct preparation of electron microscope grids. There fore ail chemical digestion residues on filters had to be low temperature ashed and refiltered prior to gnd preparation.The resulting grids were suitable for electron microscope analysis and generally contained a slightly greater sample density than grids prepared by LTA without chemical digestion. The BA ITER MAN AND COOK .MINERAL FIBERS IN FISH 955 measurements of fiber concentrations, and X-ray diffraction measurements of the amphibole mineral content of suspended solids and bottafi- sediments (Cook et al 1976) -- chemical digestion techniques were not chosen for this study, however, because membrane filters when ashed with the sample were found to add significant numbers of chrysottle fibers. Amphibole asbestos (crocidolite) fibers have also been found in one batch of membrane filters. Since the chemical digestion techniques also involved considerably more manip ulation of the sample without greatly increasing, analytical sensitivity, fish tissues were processed with the freezedrying/LTA method. Average sample sizes processed were 1.0 g for muscle tissue and 0.3 g for kidney and liver tissue. Thus electron microscope identification of particles in a typi cal gnd opening of 9500 pm2 involved 0 00004 g of muscle tissue and required -- 1 h to complete Fiber concentrations for Lake Superior waters were esti mated from monitoring data accumulated for different locations primarily in western Lake Superior since 1974 Only measurements made with the carbon-coated Nuclepore filter Jaffe Wtck method (Cook et al. 1976, Anderson and Long 1980) were used. Chrysottle fiber concentrations for the Deception Bay, Quebec area, were used as reported (Boulva and LeBeau personal communication). Results Minimum detectable particle concentrations are smallest where the amount of tissue prepared and the number of fields examined were largest The poor precision for concentrations determined when only a few fibers arc identified is reflector by the large range of the 95% confidence interval calculate? on the basis of a Poisson distribution for counted particles (Hailenbeck et al. 1977) Blank samples did not contain detectable amphibole fibers For the few blank samples con taining chrysottle fibers the blank concentration is calculated on the basis of the weight of the corresponding tissue sample per grid opening. Thus, the compansion of fiber concen trations in tissue and blank samples is on the basis of fibers observed in each per unit area of the grid Fish with a lifetime exposure to Lake Superior water were studied primarily for the presence of amphibole fibers. The amphibole fiber distribution pattern in Lake Superior (Fig. 2) results from counterclockwise circulation of fine (aconite tail ings discharged 11 km northeast of Split Rock. Minnesota, dunng the period of 1956-80 (Cook et ai, 1974) Lake trout, Salvehnus namaycush. captured at Split Rock are thought to have a much higher exposure to amphibole fibers than lake trout captured at Huron Bay, Michigan, far from the area of extreme amphibole fiber contamination. The Split Rock lake trout are likely to have occasionally migrated out of the area of high amphibole fiber concentrations, particularly by mov ing to the northeast along the north shore of Lake Superior. Amphibole fibers provide a good test of fiber accumulation by fish because they are limited in distribution in natural BATTERMAN AND cook MINERAL ! :beks in fish 95 7 STOOL4 I 92 Fig 3 [zon-magnesium-caicium ternary diagrams lor the major ration compositions of 100 amphibole fibers in both Lake Superior water and kidney tissue from lake irout exposed to the water Plotted ration X-ray emission intensity ratio are calculated directly from integrated KL, peak intensities obtained after background subtraction from the energy-dispersive X-1a> fluorescence spectra of individual amphtbole fibers wafers and are not likely to present a contamination problem during analysts as arc chrysouIc fibers No amphtbole fibers were tound in any of the blank samples analyzed. Large numbers of amphtbole fibers were identified in kidney tissue of the lake trout, bul only a few amphtbole fibers were identi fied in lake trout muscle and liver tissues (Table 2). Although no difference in amphibole fiber concentration could be observed in lake trout muscle tissue for Split Rock versus Huron Bay fish, large differences existed for kidr.ev and liver ussues. Only one amphtbole fiber was found in a11 the tissue examined from catfish and brook trout raised in the labora tory. but only -ft as much kidney tissue was examined for these fish when compared with the lake trout. A check of the association between amphibolc fibers in Lake Superior waier and those in ihe lake trout tissue is provided by comparison of the elemental composition as determined by EDX analysis of individual amphibole fibers in the water and fish tissue. The unique amphibole mineral particle distribution found in Lake Superior water is also found m the fish. The major canon (iron-magnesiumcalcium) compositions of 100 amphibole fibers observed both m Split Rock lake trout kidney (issue and in (he lake water indicated similar amphibole mineral fiber mixtures of -- 75% cummingtomte-gnmente [(Fe.MgfiSi.O^tOHh] and 20% actmohte [CaafFe.MgfiSisOahOHfi] (see Fig. 3). The absence of aluminum in almost all amphibolc particles identified precludes the presence of hornblende, a more com mon amphibole constituent of natural sediments and soils. The total of 15 amphibole fibers identified in Huron Bay lake trout samples fit the same amphibole mineral distribution pattern although too few fibers were identified to provide as strong an association. Arctic char. Sahelmus alpinus, were collected from two locations near the mouth of the Deception River at Deception Bay. an arm of Hudson Bay in Quebec. Canada. Chrysoltle fiber coneemrations as high as 670 million fibers per Lire in [he river mouth (Boulva and LcBeau personal communica tion) have been measured and arc attributable to a chrysotilc mining operation upstream and loading of asbestos onto ships at the bay. The anadromoas nature of the Arctic char and the dependence of water fiber concentrations on asbestos mining and shipping operations make it very unlikely that the fish were exposed to average chrysotilc fiber concentrations exceeding 10* fiberS/L in the year preceding their capiurc This uncertain exposure is reflected in the estimated average water exposure concentration (Table 1) Muscle tissue samples from all four Arctic char analyzed contained chrysolite fibers but more chrysotilc fibers were found in blank samples (Table 2) Higher chrysotilc concen trations are reported for kidney tissue, but these concen trations are only slightly greater than corresponding blank sample concentrations. Only one amphibolc fiber was found in all the Arctic char samples examined. Diatom frustule fragments were particularly common in kidney and liver tissue of lake trout and kidney tissue of Arctic char (no liver analyzed I None was observed in fish raised in the laboratory. A conservative estimate of frustule fragments was made by counting each particle with a porous, diatom frustuie-hke appearance All fragments analyzed by EDX gave spectra containing only a silicon K,, peak as expected for diatom frustules which arc composed of amorphous silica The concentrations in Split Rock and Huron Bay lake trout kidney tissue were 100 and 200 particles/mg. respectively. The identification criteria prevented counting many small fragments that were undoubtedly present in the samples Little comparison of amphibole fiber size distributions is possible since only lake trout kidney tissue contained enough fibers to allow a good measure of the incidence of different 958 CAN ; FISH AQUAt SCI . VOL 38. 1981 sizes of particles Typical fiber size distributions .n any media are characterized by large percentages of small particles The mean amphiboie fiber length and width for Split Rock laJcc trout kidney tissues were 1.99 and 0.37 pirn as compared to 1 50 and 0.28 jam for Huron Bay lake trout kidney tissue. Lake trout muscle tissue amphiboie fibers (only seven identi fied) had a mean length of 1 06 pim and a mean width of 0.22 wm. The kidney preparations were often ___ y with debris, so many short or thin fibers may not have been observed, and this affects both fiber concentrations and size distribution. Mean amphiboie fiber lengths and widths typi cal for western Lake Superior wafer are 2.0 and 0 3 pm fCooket ai. 19761; storm rcsuspension of settled particles and proximity to the tailings discharge may raise these values Amphiboie fibers transported to other parts of Lake Superior are expected to be primarily smaller panicles Discussion The primary purpose for this investigation was to determine if freshwater fish can accumulate asbestos and analogous mineral fibers and thereby act as a vehicle for human inges tion. Although only seven amphiboie fibers identical with those in Lake Superior water were found in Lake Superior lake trout muscle tissue, we conclude that trace concentrations of small fibers can be present in fillets from fish exposed to mineral fibers. The concentration in muscle tissue appears to be on the order of 10the water exposure concentration (expressed as fibers per gram). Thus the human ingestion of fish from asbestos-contaminated waters is unlikely to be a concern as in the case of drinking water supplies. The greater number of amphiboie fibers in lake trout kidney tissue may be related to the large volume of blood flowing through that organ. The possibility that the urinary tract serves as an elimination route for these small particles as demon strated for mammals (Cook and Olson 1979) has not been investigated. The 36-fold difference in amphiboie fiber con centrations for Split Rock and Huron Bay lake trout kidney tissue parallels the great difference in water fiber concen trations at the two locations The tendency for lake trout found at Split Rock to migrate to areas with lower fiber concen trations and lake trout found at Huron Bay to migrate to areas with higher fiber concentrations may explain why the tissue fiber concentration difference is not even greater Another factor may be the drift of food organisms into the study areas from areas of higher or lower amphiboie fiber water concentrations. The absence of amphiboie fibers in the catfish and brook trout raised in amphiboie fiber-contaminated water on an arti ficial diet suggests th3f ingestion of food organisms contain ing the fibers is the primary route for fiber accumulation in the lake trout. If so. this study provides indirect evidence for the small volume of water normally ingested by freshwater fish. We intend to expose trout to a well-measured dose of asbestos in food and compare the amount accumulated in tissue to amounts found tn fish exposed only through contaminated water The possible passage for microscopic mineral fibers through the gastrointestinal mucosa of fish and subsequent hematogenous or lymphatic transport to other tissues, particu larly the kidney, is supported by reports of such a mechanism m mammals such as mice (LcFcvre et al. 1978), rats (Sanders and Ashworth 1961sheep (Nottle 1977). baboons (PatelMandlik ct aI. 1979), and man (Cook and Olson 1979). In a study of tissues of a neonate baboon that ingested asbestos, the highest fiber concentrations were found in the kidney (Patel-Mandlik 1980), A possible alternate pathway for liber accumulation in fish, passage through gtll membranes, seems to be of minimal significance since fibers were not found in fish exposed only to libera in water The pattern of diatom frustuie fragment occurrences in tissues also supports inges tion as the primary route for particle accumulation Two analytical problems require further work. Techniquesmust be perfected which reduce chrysotile fiber contamina tion during tissue preparation to levels which do not interfere with the ability to detect chrysotile fibers in the tissue. The apparent association in this study of chrysotile fibers m blanksamples with tissue samples having the highest concentrations of chrysotile may be due to cross contamination from the tissue tube to the clank tube in the low temperature ashing step. Amphiboie fibers which are more massive than chrysoule fibers were not found in blank samples of tissues from unexposed fish. Also needed is the development of even more sensitive tissue digestion techniques to allow faster elec tron microscope examination of more tissue residue. Thus detection limits could be lowered and concentrations such as those reported for muscle tissue in this study could be deter mined with greater precision Acknowledgments The assistance of Vince Mattson and Anthony Carlson of this laboratory, and lean Boulva, Fisheries and Oceans, Government of Canada, in obtaining the fish sample, is greatly apppreciaied The laboratory assistance from Diane Huseby, Diana Longne. and Mike Kline in preparing samples is also greatly appreciated Drawings were OO prepared by Barbara Halhgan of this laboratory and ihc photographic --f work by Doug Lothcnbach and Vitkt Land of the University of f--^ Minnesota, Duiuih ^ -P~ Anderson, C. H . and J M Lono 1980 interim method for detcr-^-. mining asbestos in water EPA-600,'4-80-005 U S Government Printing Office, Washington, D C 34 p Berkley, c , J Churo. I J Selikoff, and W E Smith 1965 TheLO detection and location of mineral fibers in tissue Ann N YCO Acad. Sci. 132 48-63 Carter, R E . and W F Taylor 1980. Identification of a partic ular amphiboie asbestos fiber m tissues of persons exposed to a high oral intake of the mineral Environ Res 2F 85-93 Cook, P M. 1979. Preparation of extrapulmonary tissues and body fluids for quantitative transmission electron microscope analysis of asbeslos and other mineral particle concentrations. Ann N Y Acad Set 330 717-724 Cook. P M.G E GUss, and 1 H Tucker 1974 Asbestiform amphiboie minerals detection and measurement of high concen trations in municipal water supplies Science 185. 853-855 Cook, P. M , and G F Olson 1979. Ingested mineral fibers elimination in human unne Science 204 195-198 Cook. P M . 1 B. Rubin, C J. Maggiore, and W / Nicholson 1976 X-ray diffraction and electron beam analysts of asbestifomi minerals m Lake Superior waters, p 1-9, Vol 2. Session 34, Paper I. Proceedings of an International Conference on Environmental Sensing and Assessment (Institute of Electrical and Electronics Engineers. New York. 1976) Haujenbeck, W H.. E. H. Chen, K. Patel-Mandlik. and A H Wolff 1977 Precision of analysts for waterborne batterman \nd cook mineral fibers in fish 959 chrysculc asbestos by tmni.mis.sion election microscopy Bui;. Environ. Conum Toxicol I? 551-558 Hatband. e 1974 Der cir.fluss von asbcsta/aJlprodukten auf die miesmuschel Wasscr. Luft Beir !8 1-3 Lee. D H K 1974 Biological effects of ingested asbestos report and cummcnurv Gnvirun Health Pcrspcct 9 113-122. LeFfare.M E..J W Vanderhoff. J A Laissue. andD D Joel. 19"8 Accumulation of 2-p.m latex pamcles-in mouse Pc ye patches during chrome latex feeding Expenentia 34 120- 125 Nottt.e.M C 1977 Plant panicles in kidneys and mesentenc lymph nodes of sheep Ails: Vet J 53 405 Patej.-Manduk, K } 1980 Distribution of orally administered ctuysotile asbestos in newbom baboon body. EPA-600/l-80022 L'S Government Printing Office. Washington. D C 15 p Patel-Mandlik. K J . W H Hallenbecx, and I R M:llette 1979 Asbestos fibers I. A modified preparation of tissue samples for analysis by electron microscopy 2 Presence of fibers in tissues of baboon fed chrysoulc asbestos. J Environ. Pathol Toxicol 2 1385-095 Sanders, e. . and C T Ashworth 1961. A study of paniculate intestinal absorption and hepatocellular uptake Exp Cell Res 22 137-145 ^6 / f?*/OOlS