Journal of Clinical InvestigationVol. 42, No. 7, 1963

THE ROLEOF ENDOTOXINDURINGTYPHOID FEVERANDTULAREMIA IN MAN. I. ACQUISITION OF TOLERANCE

TO ENDOTOXIN*

By SHELDONE. GREISMAN,t RICHARDB. HORNICK, FRANKA. CAROZZA, JR.,AND THEODOREE. WOODWARD

(From the Departments of Medicine and Physiology, University of Maryland School of Medi-dicine, Baltimore, Md.)

(Submitted for publication January 16, 1963; accepted March 7, 1963)

Endotoxins prepared from gram-negative bac-teria induce profound physiologic derangementsin man (1), yet their clinical significance duringillnesses induced by gram-negative bacteria isspeculative. One major objection to ascribing anysignificant role to endotoxin in mediating thepathophysiologic changes of infection is based onthe phenomenon of acquired tolerance. When-ever endotoxin is administered daily to man assingle iv injections, resistance to the toxic and py-rogenic activities develops rapidly (2). Such ac-quisition of tolerance suggested to some investiga-tors that circulating endotoxin plays no majorrole in those infections in which fever is sustained(3, 4). Nevertheless, tolerance to Brucella endo-

toxin is not acquired during brucellosis, a gram-negative infection occasionally characterized by asustained febrile course. Indeed, since the infectedhuman host hyperreacts to this material, Spinksuggested in 1954 that Brucella endotoxin con-tributes significantly to pathogenesis (5).

Endotoxin has also been regarded as a primepathogenic factor in typhoid fever. In 1898, Mal-lory concluded, "The typhoid bacillus produces amild diffusible toxine, partly within the intestinaltract, partly within the blood and organs of thebody. This toxine produces proliferation of en'dothelial cells which acquire for a certain lengthof time malignant properties. The new formedcells are epithelioid in character . . . and arecharacterized by marked phagocytic properties

* This study was supported by the Commission on Epi-demiological Survey of the Armed Forces EpidemiologicalBoard and the International Center for Medical Re-search and Training Program, U. S. Public HealthService research grant E 4162.

t Recipient of Career Development Award from theNational Institutes of Health under contract HE-K3-15,237-Cl.

(6)." Subsequent data suggest that Salmonellatyphosa endotoxin may induce other features oftyphoid illness. This evidence, based largely uponthe striking clinical similarities of host alterationsduring typhoid fever and those observed after ivadministration of purified S. typhosa endotoxin,has been reviewed recently (7). This reviewstresses the circumstantial nature of clinical evi-dence and the need for definitive data on the roleof endotoxin.

Since 1959, we have attempted to detect physio-logically active quantities of circulating endo-toxin during gram-negative bacterial infections inman. These investigations were initiated aftersuccessful development of a reproducible model oftyphoid fever in volunteer subjects during studiesevaluating the efficacy of typhoid vaccines. Ob-servations of a variety of physiologic parametersin the same subject before, during, and after in-fection provided a unique opportunity to examinethe evidence that endotoxin contributes to patho-genesis. Since the phenomenon of tolerancerepresents a characteristic and reproducible re-sponse to repeated exposure to circulating endo-toxin, the initial studies were conducted to de-termine whether tolerance is acquired followingtyphoid fever. Parallel studies were carried outduring infection induced with Pasteurella tularen-sis, since this agent possesses endotoxin and pro-duces clinical manifestations similar to typhoidalinfections. Control studies were performed em-ploying the agent of sandfly fever, which pro-vokes an infectious disease in which endotoxinpresumably does not participate.

MATERIALS AND METHODS

Volunteers for these studies were male inmates of theMaryland House of Correction, Jessup, Md., whose agesranged from 22 to 52. Each volunteer was fully ap-

1064

ENDOTOXINDURINGTYPHOID FEVERAND TULAREMIA

W.M. 39 WM.

.___

CHLORAMPHENICOL- 4 5 day course 4

2 4gm. stat. 2gm.every 6hrs.

0 314' 5T 6 ' 7 ' 8 9 '10' 11 ' 12' 13' 14' 15' 16' 17 18 19'20' 21'22' 3' 24 25'26'27

DAY POSTINFECTION

BLOOD OO+ + + + + + 00 0CultureS.typhoso

STOOL + + + O + + + ++000+ 0 0 0 0 0 0 00 0 0

MARROWAgglutinations 0 40 40 1280 1280 1280 1280 1280

S. typhoso H 40 80 640 2560 2560 5120 5120

[WBC(thousands) WBG 10.9 1LO 11.85 9.655.4 57 5.5 4.55 3.5 5.75 4.1 9.55 9.05 8.0 6.05 5.1

FIG. 1. TYPICAL PATTERN OF THE FEBRILE COURSEOF INDUCED TYPHOID FEVER. Viable Salmotnella typhosa adminis-tered orally at time indicated by arrow.

prised of all aspects of the investigation and the at-tendant risk. They were free to cease participation atany time. Complete medical surveys were performedto verify the fitness of each participant.

Infectious agents and nature of clinical illness.

Typhoid fever. Details of the preparation and adminis-tration of the inocula of S. typhosa used to induce ty-phoidal infection will be presented elsewhere (8). The"Quailes" strain of S. typhosa, which is a phage D-1 typecontaining Vi antigen but which is 0 agglutinable was

used. Varying infecting doses were employed. Illnesswas regularly produced following ingestion of 1 X 10 to1 X 109 organisms. Figure 1 depicts the febrile course

in a typical subject. The onset of fever was generally ac-

companied by headache, abdominal pain, arthralgia, ano-

rexia, and malaise. Blood and stools were culturedfrequently to aid in verification and control of the dis-ease. After an average of 48 hours of illness, each pa-

tient was treated with chloramphenicol,l 3 to 4 g dailyfor 5 days. After one week without drug, an additional3 g of chloramphenicol daily for 5 days completed theusual therapeutic course.

Tularemia. Tularemia was induced either aerogeni-cally 2 or by intradermal inoculation. Strain SCHU-S4

1 Chloromycetin, generously supplied by Parke, Davis& Co., Detroit, Mich.

2Aerosol exposures were performed by Mr. WilliamGriffith and associates, Aerobiology Division, FortDetrick, Md.

of P. tularensis 3 was used in all instances. Completedata will be given elsewhere (9). Volunteers receivedan intradermal injection of 1,000 human infectious doses.After 24 hours, a faint localized erythematous maculeappeared, which in 48 hours became an indurated papule2 cm in diameter. The erythema extended although thepapule remained unchanged. Fever developed about day3 accompanied by systemic symptoms, predominantlyheadache and chilliness. Central necrosis of the papuleusually occurred on day 4, followed by formation ofa black eschar. Lymph nodes in the epitrochlear andaxillary areas draining the lesion enlarged on day 5 andcontinued to enlarge rapidly during the subsequent 2 days.After 36 hours of fever, streptomycin therapy was ini-tiated in a dose of 2 g daily intramuscularly. This pro-

duced dramatic relief of symptoms, lysis of fever, andgradual healing of the local lesion with regression of theregional lymph nodes.

Resipartory tularemia became clinically apparent aftera 3- to 5-day incubation period, with fever, headache,substernal pain, nonproductive, hacking cough, and back-ache. Streptomycin, employed in most infected volun-teers, controlled the infection promptly.

Sandfly fever. Sandfly fever was induced by iv ad-ministration of plasma collected during the acute phaseof disease from previously infected volunteers.4 Plasma

3 Kindly prepared and supplied by Dr. Henry Eigels-bach, Microbiology Division, Fort Detrick, Md.

4 The original stock plasma containing the Sicilianstrain of this virus was kindly supplied by Dr. BenjaminSweet.

TEMPERATURE0F

1065

GREISMAN, HORNICK, CAROZZA, AND WOODWARD

for volunteers was preserved at - 700 C and thawed im-mediately before administration. After an incubation pe-riod of 3 days, illness was heralded by sudden rise intemperature, frontal headache, retroorbital pain, backache,myalgia, and anorexia. Physical findings included vesicleson the soft palate, conjunctival injection, and erythema-tous flushing of the face and neck during the febrile pe-riod. Fever usually persisted for 2 to 3 days, and as thefever waned, clinical improvement rapidly developed.

Bacterial endotoxinsTwo preparations of bacterial endotoxin were em-

ployed for assay of endotoxin tolerance, a highly purifiedendotoxin of S. typhosa (0-282),5 previously described(10), and Escherichia coli endotoxin (Difco, 0127B8),6prepared by a modified Boivin method (11). Both endo-toxin preparations were suspended in sterile, pyrogen-free, physiologic saline and stored in rubber-stopperedvaccine bottles at 40 C. The S. typhosa endotoxin wasdiluted to a concentration of 0.5 /Ag per ml and the E. coliendotoxin to 1.0 ltg per ml. The same bottle of dilutedendotoxin was employed throughout any given study.All syringes and glassware were heated overnight in adry-air oven at 1800 C to eliminate extraneous pyrogens.The diluted endotoxin preparations were bacteriologicallysterile. Potency of the diluted endotoxins was tested

5 Kindly supplied by Dr. Maurice Landy, National In-stitutes of Health, Bethesda, Md.

6 Difco Laboratories, Detroit, Mich.

1027- T'-101-U.0

wD 100 -

4

wO 99wI-

4 98-

cI I

prior to the human studies by a standard pyrogen assayprocedure employing acclimatized, loosely restrained,albino rabbits. Repeat assays conducted at the conclu-sion of the studies, in groups of four rabbits each, withtwo dose levels for each group (0.05 and 0.005 ,sg perkilogram rabbit weight), indicated no loss of pyrogenicity.

Pyrogen assay in man

Volunteers were hospitalized before the induction oftyphoid fever, tularemia, or sandfly fever. All responsesto endotoxin were assayed after the subjects were con-fined to bed and covered with a light blanket. Eachtest was begun between 8 and 9 a.m. Flexible thermo-couples were inserted into the rectum for fixed distances(6 inches), and temperatures recorded by a Telether-mometer.7 Temperatures were monitored immediatelybefore and every half-hour after the iv injection of endo-toxin. The results were plotted on standard graph pa-per, and the area under the 7-hour fever curve, with theinitial temperature as the base line, was calculated bysumming the number of enclosed small squares. Thisvalue, termed the fever index, represents a measure of theheight and duration of the febrile response (12). Theordinates were plotted so that 100 squares represented afever index of 100 and reflected a 10 F rise in rectaltemperature for a 1-hour duration. The diurnal variationof normal body temperature was so great that tempera-

7Supplied by Yellow Springs Instrument Co., YellowSprings, Ohio.

TYPHOID STUDYTF * 21 AGE 29

FEVERINDEX

K 687K 282

I-

97+S.TYPHOSAENDOTOXIN0.5 ji9. I.V

It

SEVERITY OF S. TYPHOSA61O06DAY SYMPTOMS AGGLUTININS

(EXPOSUREDAY 7/8/60)

CONTROL 2+ 1:40(7/7/60) 2+ 1:40

8...... TH AFEBRILE DAY + 1:1280(8/4/60)__

I 2 3 4 S 6 7TI ME (HOURS)

FIG. 2. FEBRILE RESPONSEPATTERN FOLLOWINGIV INJECTION OF S. typhosa ENDOTOXIN BEFOREAND ON CONVALESCENTDAY 8 AFTER INDUCED TYPHOID FEVER.

1066

ENDOTOXINDURINGTYPHOID FEVER AND TULAREMIIA

102"TTYPHOID STUDY

TF# 24 AGE22

6f

I 2 3 4 5 6 7TI ME (HOURS)

FIG. 3. FEBRILE RESPONSEPATTERN FOLLOWINGIV INJECTION OF S. typhosa ENDOTOXINBEFORE

AND ON CONVALESCENTDAYS 3 AND 18 AFTER INDUCED TYPHOID FEVER.

TYPHOID STUDY

TF.#14 AGE 49

0I I-0

w

IL

-i49Im-

0

FEVERINDEX

750

S 255I180

TIME (HOURS)

FIG. 4. FEBRILE RESPONSEPATTERN FOLLOWING IV INJECTION OF S. typhosa ENDO-

TOXIN 12 DAYS AND 10 MONTHSAFTER CONVALESCENCEFROM INDUCED TYPHOID FEVER,AND 5 DAYS AFTER CONVALESCENCEFROMA SECONDCOURSEOF INDUCED TYPHOID FEVER.

1067

;:; 1010-0

1, 100-w

4fo

w0. 99.'I

4 980w

97-

114-

GREISMAN, HORNICK, CAROZZA, AND WOODWARD

ture seldom returned to preinj ection values. In suchinstances, the fever curves were extrapolated to the baseline.

After control responses to pyrogen were obtained,each subj ect was infected with one of the viable in-fectious agents. Responses to injection of the same en-dotoxin were reassayed at various periods during con-valescence. In addition to the fever index, subj ectivereactions (headache, chills, myalgia, anorexia) followingendotoxin administration were graded as follows: 1+ =minimal, 2+ = moderate, 3+ = severe, and 4+ = extremelysevere.

RESULTS

Acquisition of tolerance to endotoxin followingtyphoid fever. Two volunteers were given singleiv injections of 0.5 lug S. typhosa endotoxin beforeinduced typhoid fever and again during convales-cence. The reaction patterns are shown in Fig-ures 2 and 3. In each case, the fever index upontesting during afebrile days 3 to 8 was less than50%o that of the initial control value. In SubjectTF (typhoid fever) no. 24 (Figure 3), repeattesting on afebrile day 18 revealed that tolerancewas waning. A third volunteer in the typhoid

02

101°0

wIo~I-cr I O ,0 -

wa. 99-w

I-j98-

I-

wcr

series was tested initially with 0.5 jug S. typhosaendotoxin on convalescent day 12 (Figure 4). Areaction of tolerance is readily apparent, since re-peat testing 10 months later revealed that thefever index had increased over fourfold. This sub-ject was subsequently rechallenged with viableS. typhosa and again developed overt typhoid fe-ver. Upon retesting with 0.5 pug S. typhosa endo-toxin on afebrile day 5, tolerance comparable tothat following the initial bout of typhoid feverwas again apparent. In general, the intensity ofthe subjective discomfort following S. typhosaendotoxin injection paralleled the height of thefever index in any given subject.

Two additional subjects were tested with singleiv injections of 1.0-ptg doses of E. coli endotoxinbefore induced typhoid fever and again duringconvalescence.8 The reaction patterns are shown

8 In the dosages employed, the E. coli and S. typhosaendotoxins possessed comparable pyrogenic potency,elicited parallel tolerance patterns upon daily iv injections,and produced marked cross-tolerance in groups of nor-mal rabbits.

TYPHOID STUDYTF#*28 AGE39

1 2 3 4 5 6 7TIME (HOURS)

FIG. 5. FEBRILE RESPONSEPATTERN FOLLOWING IV INJECTION OF Escherichia coli ENDO-TOXIN BEFORE AND ON DAY 3 AND DURING WEEKS3 TO 6 OF CONVALESCENCEFROM INDUCEDTYPHOID FEVER.

1068

97--

1

ENDOTOXINDURINGTYPHOID FEVER AND TULAREMIA

(4+)

II IN

I *

(2+)g %

XI

I 1. %I

I ''.

x %

DSEVERITYOF S.TYPHOSA O0"DAY SYMPTOMS AGGLUTININScvbfeaine IAv 91-5A.M l_

2 4 6 8 10

TIME (HOURS)12 1 4 16 is

FIG. 6. FEBRILE RESPONSEPATTERNFOLLOWINGIV INJECTION OF Escherichia coli ENDO-

TOXIN BEFOREAND ON CONVALESCENTDAY 3 AFTER INDUCED TYPHOID FEVER.

in Figures 5 and 6. The reactions simulated thoseresulting from S. typhosa endotoxin, and the feverindexes during afebrile day 3 of convalescencewere less than 50% of the initial control values.

Repeat testing in Subject TF no. 28 (Figure 5)revealed loss of pyrogenic tolerance as convales-cence advanced, and after 12 weeks of convales-cence (not shown in Figure 5), the fever index had

TABLE I

Effect of induced typhoid fever ( TF) on responsiveness to iv administration of 0.5 jig Salmonellatyphosa endotoxin and 1.0 jig Escherichia coli endotoxin

Acquiredtolerance

Control phase Convalescent phase d1Reduction

Subject Fever Severity of Fever Severity of in feverTF no. Age index symptomst index symptomst index

yrs %14 49 750* 3 + 180 i 76

255§ 0 6621 29 687* 2+ 282 i 5924 22 1, 129* 3+ 77 i 9326 36 1,426t 3+ 704 2+ 5128 39 1,564t 2+ 462 3 + 70

Mean 69-SE 6.1

* S. typhosa endotoxin employed.t E. coli endotoxin employed.$Subjective reactions after endotoxin administration were graded as follows: 1+ = minimal, 2+ = moderate

3+ = severe, and 4 + = extremely severe.§ Second course of typhoid fever.11 Afebrile days 3 to 12,

TYPHOID STUDYTF # 26 AGE 36104

10

1069

0910-_

Idl,10

99

101

n

cdw

I1R

97-E COLI

ENDOTOXIN1.0 I.V.

T

IEArIPOURE UDA (12/UIJNO OVERTDISEASE 3+ 1:80(12/26/60)

___ 3RD AFEBRILE DAY 2+ 1:80(I/IS/SI)

I

GREISMAN, HORNICK, CAROZZA, AND WOODWARD

increased to 1,256 and the subjective reaction to2+. Subject TF no. 26 (Figure 6) manifested abiphasic febrile reaction to endotoxin before ty-phoid illness and again during convalescence. Al-though the acquisition of tolerance during con-valescence is apparent for the initial 7-hour febrileresponse, no tolerance is evident to the seconddelayed response. The explanation for this sec-ond febrile peak is unknown. As with the re-sponses to S. typhosa endotoxin, the intensity ofthe subjective reactions generally paralleled theintensity of the febrile responses. The responsesto endotoxin are summarized in Table I, and thedegree of acquired tolerance is expressed as thepercentage of reduction in fever index from initialcontrol levels. An over-all reduction of feverindex by 69% was observed following typhoidfever; 76% reduction occurred with S. typhosaendotoxin and 60%o reduction with E. coli endo-toxin.

Significance of acquired tolerance to endotoxinfollowing typhoid fever. The acquisition of toler-ance following typhoid fever, of the degree justdescribed, could not be attributed to residual ef-

fects of the initial control injection of endotoxinfor the following reasons. 1) Two subjects werechallenged with viable S. typhosa, but did not ac-quire overt typhoid illness. The reaction patternsto endotoxin in one of these subjects are shownin Figure 7. In neither subject was the fever in-dex upon second testing below 10% of the initialcontrol value. Indeed, Subject TF no. 27 (Fig-ure 7) developed S. typhosa bacteremia and yetexhibited neither overt illness nor tolerance.Moreover, this subject manifested virtually iden-tical febrile response patterns following repeatediv administration of single injections of 1.0-pugquantities of E. coli endotoxin at intervals of sev-eral weeks. Only after 80 mg of prednisone wasgiven during 16 hours before challenge was thefever index reduced below 50%o of the control value.2) In the subjects shown in Figures 3 and 5, tol-erance waned despite the repeated endotoxin ad-ministration. 3) In Subject TF no. 14 (Figure4), tolerance was apparent when this subject wasinitially tested during the convalescent phase oftyphoid fever. 4) Seven normal control subjectswere given single iv injections of S. typhosa en-

TYPHOID STUDYTF#27 AGE 39

LL0

w . FEVER-Il . I NDEX

0 **/x* -X --- 8

SEVERITY OF S.TYPHOSA"O"91~ DAY SYMPTOMS ASSLUTININSI-. EXPOSUREDAY 12//O/60O - O~~~~~~~~ONTROL(12/13/60) 2+ 1:160

W E.NCOLI NO OVERTDISEASE i I 10oAx ENDOTOXIN

..... (19/29/S0)91.0 JJQ.IV NO OVERTDISEASE ±:+I @ I(2@ /6 I -I 1 .

I 2 3 4 5 6 7TIME (HOURS)

FIG. 7. FEBRILE RESPONSEPATTERN FOLLOWINGIV INJECTION OF Escherichia coli ENDOTOXIN

BEFOREAND AFTER CHALLENGEWITH 10' VIABLE Salmonella typhosa. This subject did not de-velop overt typhoid illness, but did exhibit demonstrable bacteremia.

1070

ENDOTOXINDURINGTYPHOID FEVERAND TULAREMIA

TABLE II

Effect in control subjects (CO) of two iv injections of 0.5 ,.g of Salmonella typhosa endotoxin spaced at 1-week intervals

Acquiredtolerance

Injection 1 Injection 2Reduction

Subject Fever Severity of Fever Severity of in feverCOno. Age index symptomst index symptomst index

yrs %1 52 1,205 2+ 1,035 1 + 142 23 913 2+ 769 1 + 163 30 814 1 + 771 1 + 54 42 650 1+ 692 1+ -65 30 948* 2+ 317 0 676 36 1,127 2 + 960 1 + 157 26 1,420 3 + 927 2 + 34Mean 21ASE 9.1

* Shaking chill and febrile reaction for two successive mornings following initial reaction to endotoxin.t See footnote to Table I.

dotoxin 1 week apart. The results are tabulatedin Table II. The mean reduction in fever indexattributable to the initial exposure to endotoxinwas 21 %. In but one subject, CO (control) no. 5,was the second fever index reduced below 50% ofthe original level.9 Four control subjects tested at

9 This subject was unique in that for two successivemornings following the initial iv injection of S. typhosaendotoxin, an abrupt rise in temperature of 1.5 to 20 Foccurred, preceded by a shaking chill and frontal head-ache identical with the reactions on the morning of theendotoxin injection.

a 14-day interval exhibited only a 10% mean re-

duction in initial fever index. After a 28- to 35-day interval, even high degrees of tolerance in-duced by repeated daily iv injections of endotoxinsare lost (2). Since the interval between endo-toxin testing in the subjects given typhoid feverranged from 12 to 112 days, with a mean of 35days, it seems highly unlikely that the degree oftolerance developing after typhoid fever resultedsimply from the initial injection of endotoxin(p < 0.005 for S. typhosa endotoxin).

TABLE III

Effect of induced tularemia (TU) on responsiveness to iv administration of 0.5 pug Salmonella typhosa endotoxin

AcquiredControl phase Convalescent phase§ tolerance

Severity ReductionSubject Fever Severity of Fever of in feverTU no. Age index symptoms: index symptomst index

yrs %11* 29 823 2+ 235 0 7214* 31 1,149 2+ 701 0 3915* 33 913 1 + 150 0 8416* 29 995 3+ 648 0 3517* 25 490 2+ 88 0 8218* 26 350 0 288 0 1819* 37 1,249 3+ 286 0 7720* 27 663 2+ 462 0 30

75t 44 928 2+ 434 0 5376t 42 703 2+ 160 0 7777t 25 886 2+ 27 0 97

Mean 60±SE 13.9

* Tularemia induced by intradermal challenge with Pasteurella tularensis.t Tularemia induced by aerosol challenge with P. tularenesis,t See footnote to Table I.§ Afebrile days 3 to 8.

1071

GREISMAN, HORNICK, CAROZZA, AND WOODWARD

TABLE IV

Effect of induced sandfly fever (SF) on responsiveness to intravenous administration of 0.5 ,ug of Salmonellatyphosa endotoxin

AcquiredConvalescent phaset tolerance

Control phaseReduction

Subject Fever Severity of Fever Severity of in feverSF no. Age index symptoms* index symptoms* index

yrs %22 38 1,872 3 + 1,050 2 + 4423 43 991 1 + 1,054 1 + -624 23 661 1 + 506 0 2325 22 1,389 3 + 1,209 2 + 13

Mean 19±:SE 13.7

* See footnote to Table I.t Afebrile days 4 to 5.

The acquisition of tolerance to endotoxin fol-lowing typhoid fever could not be attributed to theantibiotic employed for therapy. Febrile re-sponses to 0.5-,ug doses of S. typhosa endotoxinwere monitored in two subjects in whom tolerancehad lapsed after convalescence from typhoid feverfor 5 and 12 months, respectively. They werethen given chloramphenicol in a dosage and man-ner identical to that previously employed duringtyphoid fever. Repeat testing with endotoxin fol-lowing completion of "therapy" revealed less than10%o decrease in fever index in both subjects.In addition, the one subject (TF no. 14) who de-veloped overt typhoid fever and did not requiretherapy acquired a high degree of tolerance toendotoxin.

From the data described it is apparent that thereductions of 50%7 or more in fever indexes, de-veloping uniformly after typhoid fever, were notattributable to the initial endotoxin testing or tochloramphenicol therapy, but represented an effectof infection per se.

Specificity of acquired tolerance to endotoxinfollowing typhoid fever. To determine whetherinfectious illnesses other than typhoid fever con-ferred tolerance in man, endotoxin responseswere monitored before and after induced tularemiaand sandfly fever. In 11 subjects given tularemia,the fever indexes and subjective reactions weremonitored after administration of 0.5 jug of S.typhosa endotoxin. The results are shown inTable III. The mean reduction in fever index onafebrile convalescent days 3 to 8 was 60% of theinitial control level. When this value is compared

with that of the normal control subjects, it is un-likely (p < 0.05) that this degree of tolerancecould be attributed to the initial testing with endo-toxin. Moreover, two other subjects failed to de-velop overt illness after attempts to infect. Whensuch subjects were tested with endotoxin at in-tervals comparable to the previous group, thefever indexes were reduced in both cases by lessthan 13%o of the initial control values.

In sharp contrast to the findings in typhoidfever and tularemia, subjects convalescing fromsandfly fever acquired no tolerance to the pyro-genic action of S. typhosa endotoxin other thanthat which could be accounted for simply by theresidual effect of the initial control testing. TableIV illustrates the fever indexes and subjectivereactions of four volunteers following the iv ad-ministration of 0.5 ,ug of S. typhosa endotoxinbefore and on afebrile days 4 to 5 following con-valescence from sandfly fever. The mean fever in-dex during the convalescent phase was reducedonly 19%o from that of the control period. Thisparalleled the mean reduction of 21 % resultingfrom similarly spaced injections of S. typhosa en-dotoxin in the normal control subjects.

DISCUSSION

Although several investigators have suggestedthat the clinical symptoms of typhoid fever may bemediated by circulating endotoxin (6, 13), fewstudies have evaluated critically those physiologicalterations that could be attributed specifically toits action. The most definitive studies deal withthe phenomenon of acquired tolerance to the pyro-

1072

ENDOTOXINDURING TYPHOID FEVER AND TULAREMIA

genic activity of endotoxin. If S. typhosa endo-toxin participates in the pathogenesis of typhoidfever, tolerance to this endotoxin should developduring the course of illness. Moreover, since en-dotoxin from one bacterial species elicits toleranceto those from other species (2, 12), the phenome-non should not be restricted to S. typhosa endo-toxin. To test this premise, Neva and Morganassayed tolerance to the pyrogenic activity of S.typhosa and Shigella dysenteriae endotoxins in 22patients during the afebrile convalescent days 4 to6 following naturally acquired typhoid fever; sig-nificant tolerance (mean fever indexes, 50%o be-low control levels) was reported (13). Toler-ance, however, was assessed by comparing thefever index obtained following infection with thatelicited in groups of control subjects. Individualvariations in response to the same endotoxinwithin such control groups were considerable, andsix of the 22 subjects convalescent from typhoidfever exhibited fever indexes higher than the meanof the control group; indeed, two subjects ex-hibited fever indexes higher than or equal to themaximal control responses. Previously, Heymanand Beeson reported that two subjects convales-cent for 2 to 3 days from naturally acquired ty-phoid fever exhibited exaggerated febrile reac-tions, i.e., manifested decreased resistance to heat-killed S. typhosa (14). The evidence indicatingdevelopment of tolerance to endotoxin during con-valescence from typhoid fever could not be con-sidered decisive.

Our findings, with each subject as his indi-vidual control, indicate unequivocally that a highdegree of tolerance to S. typhosa and E. coli en-dotoxins is acquired during the convalescent phaseof typhoid fever in man. Further studies are re-quired to define specific differences in the degreeof acquired tolerance to these homologous andheterologous endotoxins. The acquisition of thislevel of tolerance was not attributable to the resid-ual effects of the initial testing with endotoxin.However, excluding the role of antibiotic (chlor-amphenicol) therapy in producing this endotoxintolerance required special consideration. Essen-tially three problems were involved: 1) antibioticsalter normal host resistance to endotoxin (15-17); 2) antibiotics administered during typhoidfever might accentuate endotoxemia and lead totolerance; and 3) shortening the course of illness

might interrupt subsequent changes in host-tolerance mechanisms. The first possibility wasvirtually excluded by finding that subjects testedlate in convalescence from typhoid fever, aftertolerance had lapsed, acquired no significant tol-erance to S. typhosa endotoxin after full dosagesof chloramphenicol. The latter two possibilitiescannot be excluded, but appear unlikely, as judgedfrom the results of Neva and Morgan in untreatedsubjects. Also, one subject with overt illness inour series required no antibiotic therapy, but devel-oped endotoxin tolerance comparable to that ofsubjects treated with chloramphenicol.

Two subjects in the present study failed tomanifest overt illness after ingestion of 106 and109 viable S. typhosa. The latter subject devel-oped demonstrable S. typhosa bacteremia. Bothsubjects responded to S. typhosa endotoxin in anormal manner before and after the administrationof viable organisms, i.e., no tolerance to endotoxinexisted before or developed after ingestion oftyphoid bacilli. From these observations, it ap-pears that the mechanisms by which man can resistovert typhoid illness are not identical to thosethat confer resistance to endotoxin.

There was a striking parallelism between thealtered endotoxin responsiveness of typhoid feverand that of tularemia, which is another illness in-duced by a bacterial species containing endotoxin 10and producing a similar clinical syndrome (20).Thus the acquisition of significant tolerance toendotoxin was demonstrable during the convales-cent phase of tularemia induced either by intra-dermal or aerosol challenge. As in typhoid pa-tients, this degree of tolerance was not ascribableto the residual effects of the initial testing withendotoxin, although the possible influence of strep-tomycin therapy remains unsettled. As in theirresponses to S. typhosa, subjects given P. tularen-sis could resist overt illness although fully capableof responding to endotoxin before and afterchallenge.

The present findings of acquired endotoxin tol-erance following recovery from human typhoid

10That P. tularensis contains endotoxin was demon-strated by isolation, with techniques similar to those usedfor preparing Shigella flexneri endotoxin (18), of afraction possessing pyrogenic, Shwartzman-phenomenon-inducing (preparatory and provocative), and lethal ac-tivity for rabbits (19).

1073

GREISMAN, HORNICK, CAROZZA, AND WOODWARD

fever and tularemia are in apparent contrast withthose reported in brucellosis. Enhanced, ratherthan tolerant, pyrogenic and subjective reactionswere observed in subjects recovered from brucel-losis when tested with Brucella endotoxin (21).The explanation for this divergence of results isuncertain, but may be related to the endotoxinpreparations employed. Whereas the iv injectionof 0.5 ug of the endotoxins here employed con-sistently induced brisk febrile reactions in normalsubjects, 100-ug doses of the Brucella endotoxinemployed by Abernathy and Spink (21) failed toelicit any febrile reactions. Moreover, the febrilereaction pattern produced by Brucella endotoxinin subjects recovered from brucellosis was mark-edly dissimilar from that produced by classical en-dotoxin preparations. The time of onset of feveraveraged 6.1 hours, and the duration of the fe-brile reaction averaged 14 hours, in contrast toapproximately 1 and 7 hours, respectively, withendotoxin preparations reported in our studiesand by most investigators. Conceivably, the ma-terial employed as Brucella endotoxin was notprimarily a pyrogen, but rather an antigen thatproduced fever on the basis of hypersensitivity inthe sensitized human host. Studies with classicalendotoxin preparations in convalescent brucellosispatients may resolve this discrepancy.

The significance of the acquisition of endotoxintolerance during the convalescent phase of typhoidfever and tularemia is linked to two major con-siderations. 1) Do these data constitute specificevidence for the participation of circulating endo-toxin during illness? 2) Does this reaction repre-sent an important defense mechanism in recoveryfrom typhoid fever [as initially postulated byNeva and Morgan (13)] and from tularemia?The first problem of endotoxin participation isdifficult to resolve. Studies of a variety of in-fections in man show that the acquisition of endo-toxin tolerance after infection develops only aftertyphoid, paratyphoid fever, and malaria (13, 14).In the current studies, no significant tolerance ap-peared after sandfly fever, a viral disease in whichendotoxin presumably is not involved. Althoughit thus appears that fever and infection per se donot provoke endotoxin tolerance in man, it wouldbe premature to conclude that the acquisition ofendotoxin tolerance following typhoid fever andtularemia is due to endotoxemia. Indeed, no

evidence is yet available that P. tularensis endo-toxin is capable of mediating any of the observedtolerance to S. typhosa endotoxin. The findingsof acquired tolerance, however, offer the strongestsuggestive evidence thus far available supportingthe concept that physiologically active quantitiesof circulating endotoxin contribute to the patho-genesis of these illnesses. The latter problem ofthe importance of endotoxin tolerance in the re-covery process is currently under investigation.

SUMMARY

With each human subject as his own control,it was shown that a high degree of tolerance to thepyrogenic action of Salmonella typhosa andEscherichia coli endotoxins is acquired by afebriledays 3 to 12 following convalescence from in-duced typhoid fever. Such tolerance is not at-tributable to the residual effects of control test-ing with endotoxin, nor to the chloramphenicolemployed for therapy. Significant tolerance toS. typhosa endotoxin developed by afebrile con-valescent days 3 to 8 following human tularemiainduced by either the intradermal or aerosol route.No significant tolerance to endotoxin developedin subjects given viable S. typhosa or Pasteurellatularensis who failed to manifest overt illness, orthose given a viral disease, sandfly fever. Suchfindings support the premise that physiologicallyactive quantities of circulating endotoxin con-tribute to the pathogenesis of typhoid fever andtularemia in man.

ACKNOWLEDGMENTS

The authors express their deepest appreciation to thevolunteers who contributed so faithfully to these studies,to the house officers of the University of Maryland Hos-pital who assisted in care of these patients, and to theofficials of the Maryland House of Correction for theirgenerous cooperation and keen interest, without whichthese studies could not have been accomplished.

REFERENCES

1. Favorite, G. O., and H. R. Morgan. Effects pro-duced by the intravenous injection in man of atoxic antigenic material derived from Eberthellatyphosa: clinical, hematological, chemical andserological studies. J. clin. Invest. 1942, 21, 589.

2. Morgan, H. R. Resistance to the action of the en-dotoxins of enteric bacilli in man. J. clin. Invest.1948, 27, 706.

1074

ENDOTOXINDURINGTYPHOID FEVERAND TULAREMIA

3. Beeson, P. B. Discussion of paper by W. WSpinkon the significance of endotoxin in brucellosis:experimental and clinical studies. Trans. Ass.Amer. Phycns 1954, 67, 292.

4. Atkins, E. Pathogenesis of fever. Physiol. Rev.1960, 40, 580.

5. Spink, W. W. The significance of endotoxin inbrucellosis: experimental and clinical studies.Trans. Ass. Amer. Phycns 1954, 67, 283.

6. Mallory, F. B. A histological study of typhoid fever.J. exp. Med., 1898, 3, 611.

7. Greisman, S. E., T. E. Woodward, R. B. Hornick,and M. J. Snyder. Role of endotoxin in typhoidfever. Bull. N. Y. Acad. Med. 1961, 37, 493.

8. Hornick, R. B., T. E. Woodward, M. J. Snyder,L. J. Morse, and F. R. McCrumb, Jr. Studies ontyphoid fever induced in volunteers. I. Clinicalmanifestations. In preparation.

9. Hornick, R. B., E. Gangarosa, J. T. Bulkeley, F. R.McCrumb, Jr., and T. E. Woodward. Studies onthe immunoprophylaxis of tularemia. In prepara-tion.

10. Webster, M. E., J. F. Sagin, M. Landy, and A. G.Johnson. Studies on the 0 antigen of SalmontellatYphosa. 1. Purification of the antigen. J. Im-munol. 1955, 74, 455.

11. Boivin, A., and L. Mesrobeanu. Recherches sur lesantigenes somatiques et sur les endotoxines desbacteries. I. Considerations generales et exposedes techniques utilisees. Rev. Immunol. (Paris)1935, 1, 553.

12. Beeson, P. B. Tolerance to bacterial pyrogens. I.Factors influencing its development. J. exp. Med.1947, 86, 29.

13. Neva, F. A., and H. R. Morgan. Tolerance to theaction of endotoxins of enteric bacilli in patientsconvalescent from typhoid and paratyphoid fevers.J. Lab. clin. Med. 1950, 35, 911.

14. Heyman, A., and P. B. Beeson. Influence of variousdisease states upon the febrile response to intra-venous injection of typhoid bacterial pyrogen. J.Lab. clin. Med. 1949, 34, 1400.

15. Miller, C. P., W. D. Hawk, and A. K. Boor. Pro-tection against bacterial endotoxins by penicillinand its impurities. Science 1948, 107, 118.

16. Miller, C. P., and A. K. Boor. Protection of miceagainst lethal action of gonococcal endotoxin bypenicillin. Proc. Soc. exp. Biol. (N. Y.) 1946, 61,18.

17. Quan, S. F., T. H. Chen, and K. F. Meyer. Pro-tective action of antibiotics against the toxin ofPasteurella pestis in mice. Proc. Soc. exp. Biol.(N. Y.) 1950, 75, 548.

18. Cluff, L. E. Immunochemical study of a bacterialendotoxin: Shigella flexneri type Z. J. exp. Med.1954, 100, 391.

19. Cluff, L. E. Personal communication.20. Cluff, L. E. Tularemia in Principles of Internal

Medicine, 4th ed., T. R. Harrison, Ed. New York,McGraw-Hill, 1962, p. 963.

21. Abernathy, R. S., and W. W. Spink. Studies withBrucella endotoxin in humans: the significance ofsusceptibility to endotoxin in the pathogenesis ofbrucellosis. J. clin. Invest. 1958, 37, 219.

1075