An analytical method for detecting TCDD (dioxin): levels of TCDD in samples from Vietnam.

2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) is an extraordinarily toxic substance that is produced as an unwanted side product in the industrial synthesis of 2,4,5-trichlorophenol, an intermediate in the manufacture of the herbicide 2,4,5-trichlorophenoxyacetic acid (2,4,5-T) (1, 2). Because of its chemical stability and its lipophilic nature, the possibility exists that TCDD released into the environment could accumulate in food chains. A direct test of the possibility of biologically significant accumulation in animal tissues requires an analytical method able to detect TCDD at levels well below those known to be toxic. The lowest value known for the lethal dose of TCDD is that observed in the guinea pig, for which the single oral dose LD50 is 600 parts per trillion (ppt) body weight (3). Allowing for sublethal toxic effects and providing for a conservative margin of safety, it seems desirable to have an analytical sensitivity of at least 1 ppt. For a 1-g sample this means the method must have a sensitivity of, about 10-'2g or 1 picogram (pg). The most common method for analyzing chlorinated organic compounds in tissue samples is gas-liquid chromatography (GLC) with an electron capture detector. Its limit

2,3,7,8-Tetrachlorodibenzo-p-dioxin (TC-DD) is an extraordinarily toxic substance that is produced as an unwanted side product in the industrial synthesis of 2,4,5-trichlorophenol, an intermediate in the manufacture of the herbicide 2,4,5-trichlorophenoxyacetic acid (2,4,5-T) (1,2). Because of its chemical stability and its lipophilic nature, the possibility exists that TCDD released into the environment could accumulate in food chains. A direct test of the possibility of biologically significant accumulation in animal tissues requires an analytical method able to detect TCDD at levels well below those known to be toxic. The lowest value known for the lethal dose of TCDD is that observed in the guinea pig, for which the single oral dose LD50 is 600 parts per trillion (ppt) body weight (3). Allowing for sublethal toxic effects and providing for a conservative margin of safety, it seems desirable to have an analytical sensitivity of at least 1 ppt. For a 1-g sample this means the method must have a sensitivity of. about 10-12g or 1 picogram (pg).
The most common method for analyzing chlorinated organic compounds in tissue samples is gas-liquid chromatography (GLC) with an electron capture detector. Its limit *Harvard University, Cambridge, Massachusetts 02138. of detection for TCDD, about 10-l0g, is inadequate. This method is also susceptible to interference from other compounds and so is not very specific.
Mass spectrometry offers better possibilities. It is high sensitive and in the high resolution mode of operation it is highly specific. We have previously described a time averaged mass spectroscopic method with an adequate limit of detection (4). However, full sensitivity could not be realized in most sample types because of interference from DDE (a major degradation product of DDT) and polychlorinated biphenyls (PCBs). In this paper we describe a cleanup procedure that overcomes this difficulty.
Homogenized samples are saponified in alcoholic potassium hydroxide and extracted with hexane. The extract is shaken with sulfuric acid and chromatographed on alumina. Elution with carbon tetrachloride-hexane removes most of the DDE and PCBs. Chlorinated dioxins are then eluted with dichloromethane-hexane. The TCDD containing fraction is further purified by preparative gasliquid chromatography and analyzed by mass spectroscopy by use of a multichannel analyzer to average successive scans.
We also report the levels of TCDD found in a limited number of samples of fish and crustaceans from locations in South Vietnam near areas heavily exposed to 2,4,5-T.
Activated alumina was Fisher A-540, activated at 1300 C for 24 hr.
The gas chromatograph was a Bendix Model 2200 equipped with a thermal conductivity detector. The column was 5%o SE-30 on 60/80 Chromosorb W, 2 m x 2 mm (id) stainless steel. The trap for preparative gas chromatography was a 150 mm x 1.5 mm (id) glass tube packed with 30 mm of glass wool.
An Associated Electrical Industries MS-9 double focusing mass spectrometer and a Varian 1024 time-averaging computer interfaced with the MS-9 as described earlier (4), were used.
Cleanup Procedure for the Analysis of TCDD in Tissue Samples (1) The sample was weighed and homogenized with 1.0-1.2 parts EtOH.
(2) This homogenate was transferred to a round-bottomed flask equipped with a reflux condenser (Teflon tape used on the ground glass joint). The sample was spiked with approximately 1000 ppt 37Cl TCDD; 2 parts 40%o aqueous KOH were added, and this mixture was refluxed for 2 hr. One part always refers to the original samples.
(3) The solution was partially cooled and 1 part hexane added.
(4) The solution was transferred to a separatory funnel, and the phases were separated. The aqueous phase was extracted with three more identical portions of hexane; the hexane extracts were combined and collected in the original round-bottomed flask.
(5) The hexane phase was transferred to the separatory funnel, the round-bottomed flask was rinsed twice with a few milliliters of EtOH and then twice with a few milliliters of hexane; the solvent was refluxed each time; and the hexane was extracted with 1 part 1.ON NaOH. (6) The hexane was extracted four times (or until acid phase was colorless) with 2 parts 95-97%o H2SO4. Emulsions were broken with a few drops of saturated Na2CO3 solution.
(7) The hexane was extracted with 1 part water, and several grams of Na20C3 were added to the hexane. (8) The hexane was filtered through a column of Na2CO, (100 mm x 10 mm id for 300 ml hexane), the Na2CO3 first being prewashed with several milliliters of hexane. (9) The hexane was concentrated to 3-4 ml (Snyder column).
(10) The hexane residue was chromatographed on a column of activated A203 (50 mm in a 5 mm disposable pipet). The column should not be prewashed. Elution was with 12 ml of 20%o CCG4 in hexane, then 1 ml of hexaane, and finally 4 ml of 20%o CH2C12 in hexane.
(11) The 20%o CH2Cl2 fraction was concentrated carefully to about 50 Fd,  ,ul benzene added, and concentration repeated to 20 FJ. (12) A few micrograms of m-terphenyl in benzene were added to the residue and the mixture subjected to preparative chromatography. The retention time of m-terphenyl relative to that of TCDD was determined beforehand and used to make certain that the TCDD collection was carried out at the right retention time.
(13) The GLC trap containing TCDD was eluted with 60,u followed by 10,ul of benzene. The total amount of eluant collected was measured, and the fraction size for the planned number of fractions (typically ten) calculated. (14) The fractions for TCDD analysis were prepared in the sample tubes described previously (4). A known amount of TCDD was added to three or more fractions for quantitation of any TCDD observed. The amount of TCDD added per fraction for Environmental Health Perspectives quantitation should be approximately three or four times the amount expected to be present. (15) The fractions were analyzed with the MS-9 instrument. Typical conditions were: source 220°C, resolution 10,000 (based on a 10% valley between peaks), trap current 1.0 mA (rhenium filament), electron multiplier 700, ionizing voltage 70 eV, time averaging at four scans per second.
(16) Peak heights were measured at m/e 321.894. The quantity of TCDD (picagrams), present in the fractions to which TCDD has not been added was computed from the ratio of their mean peak heights to the mean peak heights found with added TCDD.
(17) Steps (14)- (16) were repeated, but 37C1 TCDD was added and peak heights were measured at m/e 327.885 in order to compute the amount of 37Cl TCDD recovered. The recovery through the complete cleanup procedure was then calculated based on the amount of 37Cl TCDD added to the sample at the beginning of the cleanup.
(18) The quantity of TCDD computed in step (15) was corrected by the recovery factor obtained in step (16) to give the final result.

Sample Collection
Freshly caught fish and crustaceans were collected in South Vietnam in August and September 1970 from local fishermen. The samples were homogenized with a meat grinder, placed in acetone-rinsed glass bottles with aluminum foil-lined caps, and immediately frozen in solid CO2. Later on the same day, samples were placed in a Linde LR-35 liquid nitrogen refrigerator where they remained until analysis. Water blanks were present in the liquid nitrogen refrigerator throughout the storage period and were analyzed with the samples. Fresh Cape Cod butterfish (Poronotus tricanthus, family Stromateidae) were obtained from a local market, homogenized, and kept at -20°C until analysis. Domestic beef livers were obtained and treated similarly.

Methodology
The mass spectra of natural and 37CI TCDD are shown in Figure 1. The most intense signal for natural TCDD occurs at m/e 321.894 (nominal m/e 322), corresponding to the isotopic isomer with one atom of 37Cl and three atoms of 3-5C. The natural abundances of the Cl isotopes are 75.53 and 24.47%o, respectively. The observed spectrum for the synthetic 37CI TCDD corresponds to an isotopic purity of 95.5%o 37CI, the same as the value claimed by the manufacturer (Oak Ridge National Laboratory) of the NaCl used in the synthesis of the labeled TCDD. The synthetic 37Cl TCDD contributes only 0.042%o as much to the peak at m/e 322 as it contributes to its most intense signal at m/e 328. The contribution at m/e 320 is even lower, by a factor of nearly 100. This allows an excess of 37Cl TCDD to be added to each sample before cleanup without interfering i- M/E PTn,TR. 92 Mon. nip+.ra ohninD rlilit-nin makes possible the measurement of picogram quantities of TCDD in samples initially containing more than a millionfold excess of DDE and PCBs. Figure 2 shows the effectiveness of this procedure. The calculation of TCDD levels described in steps (14)- (16) of the experimental section assumes a linear relationship between peak height and amount of TCDD present in any given sample. Figure 3 demonstrates that the response is indeed linear over the full range of TCDD amounts introduced into the MS-9 in the course of the analyses re-B. ported here.
The reproducibility and overall recovery of the complete analytical procedure is illustrated in Table 1. A sample of beef liver was homogenized and divided into three portions each of which was then spiked with 20 ppt TCDD and 1000 ppt 37Cl TCDD. The three samples were independently put through the A. cleanup procedure up to the GLC step. Each sample was then split into three portions before preparative CLC and mass spectrometric analysis, giving rise to a total of nine separate values for the recovery of both '-I TCDD and 37Cl TCDD. The average re- 1.800 covery was 34 ± 7% for TCDD and 27 ± 5% for 37Cl TCDD. When the slight backnf nnr. ground signal at m/e 322 in an unspiked r Xl:UM Q. ITIUMSZSPVXVrU 151UWllls rrUu>U%;WU1VI"" and PCB levels in fish residue by means of alumina chromatography. Following the sulfuric acid cleanup step, the residue in hexane is added to a column of activated alumina: (A) Trace from the material eluted by 20% CH2C12 in hexane after the column was first eluted with 20% CCl. in hexane; (B) trace obtained from a similar 20% CH2Cl2in-hexane elution after the column was first eluted with 1% CH2Cl2 in hexane. Elution with 1% CH2012 in hexane was reported to be effective in reducing the amount of PCB residues (5). Elution with 20% CCL4 is clearly even more effective and was routinely used in obtaining the results reported here. with analysis of natural TCDD at m/e 322 and 320. The addition of 37Cl TCDD provides a carrier and makes possible the calculation of absolute recoveries.
An alumina chromatography step has been developed which, when combined with the cleanup steps described previously, (4)   Mean recovery 34 ± 7.2 27 ± 5.0 for A, B, and C sample of the same liver is taken into account, the calculated recoveries from the spiked samples become even more nearly equal. Experiments performed separately with each individual cleanup step established that the step with lowest recovery is preparative gas-liquid chromatography.
We conclude from these and other controls that the present analytical method provides the sensitivity and reproducibility required for biologically meaningful analyses of animal tissue samples. The method makes possible investigations of such, samples at levels approximately 10-4 times those reported heretofore (6).

Observed TCDD Levels
Signals at m/e 320 and 322 were conspicuously present in each of the fish and crustacean samples from Vietnam. The calculated levels of TCDD, summarized in Table   2, range from 18 ppt to 814 ppt, based on total wet body weight.
No peak was observed at m/e 320 or 322 with Cape Cod butterfish. The background signal corresponded to a level of 3 ppt of TCDD. No peaks were observed in water blank samples present in the liquid nitrogen refrigerator throughout the sample collection and storage period.
Confirmation that peaks observed at m/e 320 and 322 are in fact produced by TCDD is routinely provided by the criteria outlined in part A of Table 3. All three of these criteria are met by the mass spectra from each of the Vietnamese samples. The additional confirmatory procedures listed in part B of Table 3 were carried out on a sample of Vietnamese fish. This sample, carp from the Dong Nai River, exhibited a mean TCDD level of 540 ppt. The mass spectrum in the region m/e 322 is shown in Figure 4. The compound observed in this fish behaved identically to TCDD in each of the three additional confirmatory tests. We consider it extraordinarily unlikely that this-compound is anything other than a tetrachlorodibenzo-p-dioxin. In contrast to the significant amounts of 2,3,7,8-tetrachlorodi- a Letters refer to sites on map in Figure 5.
b Roman numerals refer to independent cleanups of different portions of the same sample. All values are corrected for recovery. 1. M+-COCl fragmentation peak has expected mass and isotopic isomer ratio 2. Percent recovery after partial photolytic decomposition equals that of '7Cl TCDD (7,8) 3. Partition coefficient between dichloromethanehexane and acetonitrile equals that of 5Cl TCDD (7). aSteps 2 and 3 of the additional procedures were carried out on the dichloromethane-hexane eluant from the alumina chromatography prior to preparative GLC.   Table 2. Areas heavily treated with 2,4,5-T before its use was ordered discontinued in

Discussion
Considering the limited number of samples we have analyzed and the fact that they were collected 21/2 yr ago, it does not seem appropriate to attempt any detailed evaluation of the possible toxicological significance of our results. Such discussion is made even more difficult by the complexity and incompleteness of the existing toxicological data. However, in order to provide perspective for such discussion, a tabulation of some of the principal toxicity data on TCDD is presented in Table 4. It may be noted that guinea pigs consuming their weight of food contaminated with TCDD at a level of 600 ppt would have ingested a quantity corresponding to the lethal dose. In contrast, a far greater quantity of TCDD is required to reach the LD50 cited for rats. The table shows that teratogenesis in the rat occurs at doses substantially lower than those required to kill. Lily endosperm, ambient concentration <200 (12) Feeding studies in monkeys show that dioxin poisoning is cumulative (13). Various levels of a toxic fat known to contain chlorodioxins were incorporated into the daily diet of Macaca mulatta monkeys. As pointed out by the investigators, the mean survival time depended inversely on the daily dose. A plot of their data (Fig. 6) (14). It is of obvious interest to survey appropriately chosen populations in South Vietnam more closely, especially if TCDD residues should be found in human tissue samples. Finally, turning from questions of environmental toxicology to the biological mechanisms of action, we note that TCDD seems to be particularly toxic to proliferating tissues, as suggested by its effects on spermatogenesis and hematopoiesis and its apparent toxicity to the intestinal epithelium (13) and the thymus (15). These indications are consistent with the effects of a mitotic poison, such as TCDD is known to be in the African blood lily (12) and possibly in Drosophtla melanogaster (16). We are led by these observations to speculate that TCDD may be able catalytically to disrupt microtubules, the subcellular elements of which spindle fibers are constructed and which are ubiquitous in their structural roles in cell extension and cell movement.

Summary
A procedure has been developed for the reliable detection of TCDD in animal tissues down to levels approaching 1 ppt. It makes use of chemical cleanup, preparative gasliquid chromatography, and analysis by timeaveraged high resolution mass spectroscopy.
A limited number of fish and crustacean samples was collected in South Vietnam in 1970 near areas heavily exposed to the herbicide 2,4,5-T. TCDD was detected in these samples at levels ranging from 18 to 810 ppt. TCDD was not detected in a sample of Cape Cod butterfish used as a control.
These results suggest that TCDD may have accumulated to biologically significant levels in food chains in some areas of South Vietnam exposed to herbicide spraying. Note added in proof: Overall recoveries have been increased to 60-80%o by replacing the GLC step with an additional A1203 column step. Details of this procedure will be described in a future publication.