[CANCER RESEARCH 48, 4409-4416, August I, 1988]

Laboratory Correlates of Adoptive Immunotherapy with Recombinant Interleukin-2and Lymphokine-activated Killer Cells in Humans1

D. H. Boldt, B. J. Mills, B. T. Gemlo, H. Holden, J. Mier, E. Paietta, J. D. McMannis, L. V. Escobedo, I. Sniecinski,A. A. Rayner, M. J. Hawkins, M. B. Atkins, N. Ciobanu, and T. M. EllisNCI Extramural IL-2/LAK Working Group, Audie L. Murphy VA Hospital and University of Texas Health Science Center at San Antonio. San Antonio, Texas 78284

ABSTRACT

Adoptive immunotherapy with interleukin 2 (II -2) and lymphokine-activated killer (LAK) cells (IL-2/LAK) is a technically demanding cancertherapy dependent upon large scale isolation and culture of lymphocytes.An important question is whether this technology can be accomplishedroutinely outside of highly specialized centers. In addition, no systematicexamination of laboratory correlates of IL-2/LAK therapy in humans hasbeen reported to date. The objectives of this report are to address twoissues relevant to IL-2/LAK therapy, (a) Can IL-2/LAK therapy beaccomplished outside of previously identified centers of expertise? (b)What are the relevant laboratory/clinical parameter correlations? Thesix institutions in the National Cancer Institute extramural trial treated83 évaluablepatients with renal cancer, malignant melanoma, or coloncancer with IL-2/LAK by a uniform protocol. Patients received 5 days ofIL-2 priming, then daily leukaphereses for 5 days starting 48 h after II.-2 to harvest cells. Mononuclear cells were isolated, then cultured in rollerbottles in 1-liter aliquots for 3 to 4 days at a cell density of 1.5 x 10'' per

ml with recombinant IL-2, 1500 units per ml. Cells were harvested andadministered to patients with additional IL-2. Administration of IL-2regularly induced lymphopenia and rebound lymphocytosis. Leukapher-esis yields and numbers of LAK cells generated in culture and reinfusedinto patients correlated directly with peak lymphocyte counts achievedby IL-2 administration. Mean mononuclear cell recovery per 5 days ofleukapheresis (±SEM)was 14.3 ±0.8 x Id"1. Average volume of cells

cultured per patient was 95 liters (range, 41 to 235). Mean yield of cellsharvested from cultures was 53%. Mean total number of LAK cellsinfused per patient was 7.6 ±0.4 x HI1"(range, 2 to 15.2 x Id"1). LAKactivity was measured in vitro by lysis of 51Cr-labeled natural killer-

resistant Daudi and fresh tumor targets. LAK effector cells regularlylysed these targets in vitro. Neither tumor reduction nor clinical toxicitycorrelated with dose or with cytolytic activity of LAK cells, or with otherlaboratory parameters including base-line lymphocyte count and IL-2-induced lymphocytosis. We conclude: (a) large quantities of LAK effectorcells with tumoricidal activity can be generated routinely at differentcenters; (b) neither in vitro LAK activity nor numbers of LAK cellsinfused were predictive of clinical efficacy or toxicity. There is a need toidentify other laboratory or clinical parameters more predictive of IL-2/LAK therapeutic efficacy or toxicity.

INTRODUCTION

Adoptive immunotherapy with IL-22 and LAK cells (IL-2/

LAK) is a form of cancer treatment uniquely dependent uponmajor laboratory support (1). This treatment involves the removal of autologous lymphocytes by lymphocytapheresis, theiractivation ex vivo by incubation with the lymphokine, IL-2, andsubsequently, their reinfusion to patients. As a consequence ofIL-2 activation, LAK cells acquire broad cytolytic activity demonstrable in vitro against both fresh and cultured tumor cells,including autologous tumors (2-4). Three major reports have

Received 1/4/88; revised 4/25/88; accepted 4/26/88.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1This project was supported by the following contracts from the National

Cancer Institute: U10CA32102: UIOCA21744; U10CA25827; UIOCA14958:and R01CA39489.

2The abbreviations used are: IL-2, interleukin 2; LAK, lymphokine-activatedkiller cells; rIL-2. recombinant IL-2; CNS, central nervous system; HBSS, Hanks'

balanced salt solution: LSM, lymphocyte separation medium; NK. natural killer.

indicated efficacy of IL-2/LAK against certain refractory human malignancies (5-7). However, none of these reports addressed in detail the laboratory component of IL-2/LAK therapy.

Because the generation of LAK cells is so critically dependentupon laboratory support, IL-2/LAK is an especially technicallydemanding regimen. Therefore there is concern about the practicality of this therapy for widespread application to the treatment of cancer, especially its utilization outside of a few highlyspecialized centers. In addition, there are no data presentlyavailable to address the questions of what clinical factors mayinfluence numbers or activity of LAK cells generated frompatients or whether numbers or activity of adoptively transferred LAK cells predicts tumor response or clinical toxicity inhuman patients. Evaluation of these kinds of data may aid inassessing the role of ex vivo LAK cell generation in the efficacyof IL-2 immunotherapy.

The National Cancer Institute Extramural IL-2/LAK Working Group has conducted confirmatory studies of IL-2/LAK inpatients with selected malignancies.3 Between April and De

cember, 1986, 93 patients were treated by this group at 6different centers according to a common clinical and laboratoryprotocol. This extensive multicenter experience has produced auniform data base by which to assess laboratory requirementsnecessary for treating large numbers of patients with LAK/IL-2 and by which to address critical questions about the relationship of LAK cells generated ex vivo with clinical parametersdefining patient response and toxicity.

MATERIALS AND METHODS

Patient Population. All patients had rnetastatic or recurrent renalcancer, malignant melanoma, or coloréela!cancer for which standardtherapy had failed or no standard effective therapy was available. Allpatients had clinically évaluabledisease. None had received any anti-cancer therapy for at least 4 wk, and all were recovered from toxicitiesof previous therapies. All patients had an Eastern Cooperative Oncology Group performance status of 0 or 1 and appropriate laboratoryevidence of normal bone marrow, renal, hepatic, and heart function.Patients with central nervous system involvement were ineligible.

Ninety-three patients were enrolled on study. Of these, 83 receivedLAK cells, had measurable disease, and therefore were évaluableforthis analysis. Of the 10 inevaluable patients, 4 received no LAK cellsbecause of life-threatening toxicity (2 with myocardial infarctions and2 with Grade III or IV pulmonary toxicity). Four other patients receivedno LAK cells for the following reasons: CNS métastases(1 patient);human immunodeficiency virus positivity (1 patient); refractory hyper-calcemia (1 patient); or sepsis (1 patient). One patient withdrew fromstudy. Finally, one inevaluable patient did receive LAK cells but wasineligible for further study because of lack of measurable disease.Characteristics of the 83 patients évaluablefor this study are given inTable 1.

3The 6 institutions comprising the National Cancer Institute Extramural IL-2/LAK Working Group are: City of Hope Medical Center, Duarte, CA; Einstein-Momefiore Medical Center, Bronx, NY; Loyola University Medical Center,Maywood, IL; Tufts-New England Medical Center, Boston, MA; University ofCalifornia-San Francisco, San Francisco, CA; and University of Texas HealthScience Center al San Antonio, San Antonio, TX.

4409

Research. on November 17, 2018. © 1988 American Association for Cancercancerres.aacrjournals.org Downloaded from

LABORATORY CORRELATES OF IL-2/LAK

Treatment. The clinical protocol is shown in Fig. 1. Patients receivedrIL-24 priming at a dose of 100,000 units/kg of body weight by i.v.

infusion every 8 h for a maximum of IS doses on Days 1 to 5. Somedoses were omitted, depending on patient tolerance, but no patientreceived fewer than 9 doses (median, 14). Following 2 days of rest,patients next underwent 4-h leukaphereses for 5 consecutive days, Days8 to 12, to obtain mononuclear cells. Leukaphereses were performedwith a continuous-flow cell separator (IBM-2997; Cobe Laboratories,Lakewood, CO) essentially as described by Rosenberg et al. (5, 8).Mondimi Inn cells were isolated and cultured as described below togenerate LAK cells. LAK cells were administered to patients by centralvenous catheter over approximately 30 min on Days 12, 13, and IS.On the afternoon of Day 12 patients received an infusion of cellsobtained on Days 8 and 9; on Day 13 they received cells collected onDay 10; and, on Day IS they received cells collected on Days 11 and12. Recombinant IL-2 again was administered during the period ofLAK infusion at a dose of 100,000 units/kg of body weight by i.v.infusion every 8 h for a maximum of 14 doses beginning immediatelyafter the first LAK cell infusion on Day 12. Doses were sometimesomitted depending on patient tolerance. Patients received 1 to 14 ill2 doses (median, 10) during this phase of the protocol.

For monitoring and to facilitate management of rIL-2 toxicities, allpatients were hospitalized in an intensive care unit both during priminginfusions of rIL-2 and also during administration of rIL-2 and LAKcells. Most patients received acetaminophen (650 mg every 4 h), indo-IIK-Ilini-in(2S mg every 6 h), and ranitidine (ISO mg every 12 h) during

treatment. Hypotension was treated with fluid replacement and liberaluse of the vasopressors, dopamine and phenylephrine.

Table 1 Patient characteristics

No.Median

ageSex

MaleFemaleRenal

cancer3253(19-66)°23

9Melanoma3138(19-64)247Colorectal

cancer2049(28-61)12

8

Prior chemo- or X-raytherapy

Performance status

12

01191320 119 11

* Numbers in parentheses, range.

IL-2

I 23456789 10

D°y I I I I I I Ill

MTWTFSSMTW

I I I

II 12 13 14 15 16 17 18 19 20

M I I I I I I I I I I I I I I I I I I

TFSSMTWTFS

L : LeukopheresisIL-2- 100,000 u/kg iv q8h.

LAK: Infusionol LAK cells, i.v.

Fig. 1. IL-2/LAK clinical protocol.

4 Highly purified recombinant human interleukin 2 from E. coli was used forin vivo and in vitro applications in Ihis study. rlL-2 was supplied by the CetusCorporation, Emeryville, CA.

Generation of LAK Cells (5, 6, 8). Patient cells obtained by a 4-hleukapheresis were delivered to the laboratory in a Fenwal transfer packat a final volume of 400 to SOOml. The cell suspension was diluted 1:4with calcium- and magnesium-free HBSS and then underlayed withI..SM (Litton Bionetics) in SO-ml conical centrifuge tubes (40-ml cellsuspension: 10 ml of LSM per tube). The LSM gradients then werecentrifuged at 900 x g for IS min. The plasma and platelet-richsupernatants were removed by vacuum suction, and then the mononuclear cell bands were recovered and pooled in 250-ml conical centrifugetubes (Corning 25350). The mononuclcar cells were washed twice with200 ml of HBSS, the cell pellets being collected by centrifugation at500 x g for 10 min after each wash. After the second wash, cells wereresuspcnded in 200 ml of HBSS and counted. In general, cell viabilityas judged by exclusion of trypan blue dye exceeded 95%. Following afinal centrifugation at 500 x #, the cells were cultured at a concentrationof 1.5 x 10' cells per ml in 2.3-liter roller bottles containing 1 liter of

complete LAK activation medium [RPMI 1640 medium containing 2HIML-glutamine, antibiotics, and 2% heat-inactivated human AB serum(M. A. Byproducts, Walkersville, MD)]. rIL-2 (Cetus Corporation,Emeryville, CA) was added to the bottles to achieve a final concentration of 1500 units per ml (1.5 x 10' units per roller bottle). Rollerbottles were tightly capped and incubated for 3 to 4 days in a 37'C

incubator with the roller bottle decks adjusted to turn the bottles at arate of 0.5 to 1.0 rpm.

To harvest cells from the roller bottle cultures after 3 to 4 days inculture, the LAK cell suspensions were centrifuged at 1500 rpm for 15min in 1000-ml centrifuge bottles (Nalgene). Cell pellets were pooledin 250-ml conical centrifuge tubes and then washed twice with HBSSwithout Ca2+, Mg2*, or phenol red. After the first wash, all cells were

pooled. After the second wash, cells were resuspended in 200 ml of0.9% NaCI solution and counted. Viability as judged by exclusion oftrypan blue generally exceeded 95%. Cells were resuspended in 200 mlof infusion medium (40 ml of 25% human serum albumin, 160 ml 0.9%NaCI solution for infusion, 75,000 units of rIL-2). Finally, the cellsuspension was Filtered through sterile 110 mesh Nytex (Lawshe, Rock-ville, MD) and transferred to a Fenwal transfer pack for infusion intothe patient.

Assay for LAK Activity (2-4). LAK activity was assayed in a standardchromium release cytotoxicity system. Target cells used included thestandard NK target cell line, K562; the NK-resistant cell line, Daudi;and fresh tumor cells. For the assay, target cells were washed and thenincubated with "Cr (200 nCi per 5 x IO6 cells) in LAK medium at37'C for 1 h with shaking. Radiolabeled cells were washed 3 times withmedium, incubated at 37*C for 30 min to allow dissociation of loosely

attached radioactivity, and then washed a final time with medium.Assays were performed in 96-well U-shaped microtiter plates in a Finalvolume of 150 n\ of LAK medium per well. Each well contained 5 x10' "Cr-labeled target cells. Effector cells were added to achieve effec-

tor ¡targetcell ratios of 80, 40, 10, and 2.5 to 1. Parallel wells containedtargets only, or targets plus 0.1 N HC1 or 2% sodium dodecyl sulfate,to determine spontaneous and maximal "Cr release, respectively. All

data points were performed in triplicate. The microtiter plates werecentrifuged at 500 rpm for 5 min and then incubated for 4 h at 37°C.

After incubation, the plates again were centrifuged, and then supernatants were harvested using a Skatron harvesting device. Supernatantswere counted in a gamma counter to determine "Cr cpm released from

target cells into the medium. Cytotoxic activity was calculated by thefollowing formula.

% of specific lysis

cpm released by effectors —cpm released spontaneously= : ; ; : - ; X 100%

maximal cpm released —cpm released spontaneously

Statistical Analyses (9). All statistical computations were performedwith the Biomedicai Computer Programs, P Series, 1981. Comparisonsbetween groups of patients were made using t tests or analysis ofvariance followed by Newman-Keul's multiple comparison procedure.Pearson's correlation coefficients were used to conduct bivariate analyses. Multivariate regression analyses, using BMDP's all possibleregression algorithm, were performed to identify demographic or clin-

4410

Research. on November 17, 2018. © 1988 American Association for Cancercancerres.aacrjournals.org Downloaded from

LABORATORY CORRELATES OF IL-2/LAK

ical parameters that might predict response to therapy, toxicities, orLAK cell activity.

RESULTS

Peripheral Lymphocyte Counts during IL-2/LAK. Mean peripheral lymphocyte counts for the group of patients receivingIL-2/LAK are illustrated in Fig. 2. As reported by others,administration of rIL-2 caused a dramatic fall in peripherallymphocyte counts detectable within 24 h of beginning rIL-2infusions (7, 10, 11). Discontinuation of rIL-2 priming wasfollowed within 36 h by a rebound lymphocytosis (Days 7 to12), the lymphocyte count peaking approximately 60 h afterthe last priming rIL-2 dose. The peak lymphocyte count following rIL-2 priming ranged from 1,980 to 17,526 (mean, 4,872)per microliter. The peak lymphocyte count achieved after rlL-2 was higher in patients who had received no prior chemo- orradiotherapy than in those who had received such treatment(mean, 5,444 versus 4,084 cells per n\; P = 0.01) (Table 2). rlL-2-induced lymphocytosis also correlated with age (Table 2).Patients less than age 50 yr achieved higher counts than thoseover age 50 (mean, 5,357 versus 4,183 cells per /¿l;P = 0.03).All patients received 9 to 15 priming doses of rIL-2. Mostsubjects (73%) received 13 or more doses. Patients in thissubgroup achieved higher peak lymphocyte counts (mean, 5,253per n\) than patients who received 12 or fewer rIL-2 doses(mean, 3,826 cells per n\). This difference was highly significantwith P = 0.003. Patients with malignant melanoma achievedhigher peak lymphocyte counts (mean, 5,561 cells per ¿jl)thanpatients with renal (mean, 4,520 cells per //I) or colon cancer(mean, 4,341 cells per /tl). However, this difference was accounted for by the younger age of melanoma patients (median,38 yr versus 53 and 49 yr for renal and colon cancer patients,respectively) and by the lower percentage of melanoma patientswho had received prior therapy (7 of 31 versus 25 of 52 forrenal and colon cancer). The level of rIL-2-induced lymphocytosis did not correlate with sex or with performance status(Table 2). By bivariate analysis, rIL-2-induced lymphocytosisdid correlate positively with patients' base-line lymphocyte

counts (r = 0.28; P = 0.02).Mult ¡variatoanalysis was used to identify factors which might

Table 2 Clinical correlates ofrlL-2-induced lymphocytosis

8-

7-

Leukopheresis

5.I 6H

IO

Q 5-

4-

Fig. 2.

2 4 6 8 10 12 14 16 18 20Day

Mean peripheral lymphocyte counts (±SEM)during IL-2/LAK ther-

ParameterAge

<50(44)°>50(31)SexM

(56)F

(19)Performance

status0(33)1(28)Disease

Renal cancer (30)Melanoma (28)Colon cancer(17)Doses

priming rIL-2<12 doses(12)>12

doses(63)Prior

chemo- or X-ray therapyYes(30)No

(44)Peak

lymphocytecount(cells/pl)5357

±394*

4183±3195033

±348

4397±3194750

±375

5260 ±6904520

±3585561 ±5704341 ±3603826

±3005253

±3434084

±3085444

±396P

=0.03P

=0.31P

=0.50P

=0.003P

= 0.01

apy.

" Numbers in parentheses, number of subjects.* Mean ±SEM.

be predictive of peak lymphocyte counts induced by rIL-2priming. History of no prior therapy was the most stronglypredictive variable (P = 0.008), whereas the number of primingrIL-2 doses was weakly predictive (P = 0.04). Other variablesincluding age, sex, and performance status were not predictivein the multivariate model.

Fig. 2 illustrates that a second cycle of lymphopenia andrebound lymphocytosis was associated with the therapeuticadministration of rIL-2 and LAK cells on protocol Days 12 to16. This second peak of rebound lymphocytosis occurred onDay 19, again approximately 60 h after the last therapeuticrIL-2 dose, but was substantially higher than the peak lymphocytosis observed after administration of priming rIL-2 (mean,7,221 versus 4,433 lymphocytes per ^1). This second reboundlymphocyte peak correlated strongly with the first reboundlymphocyte peak (P —0.0006), but did not correlate with thebase-line lymphocyte count (P = 0.07). A strong correlationalso was identified between the second rebound lymphocytepeak and the total number of LAK cells infused, but not withthe number of doses of rIL-2 administered with the LAK cellinfusions (data not shown).

Ex Vivo LAK Cell Generation. Leukapheresis was scheduledto coincide with the lymphocytosis induced by administrationof priming rIL-2 to maximize yields of cells for culture (Fig.2). Numbers of mononuclear cells actually isolated and put intoculture on each of the 5 days of leukapheresis are shown in Fig.3. The general pattern mirrors that of the peripheral lymphocytecount with progressive decreases in numbers of cells culturedover the 5-day leukapheresis period. Predictably, a strong positive correlation was established between peak peripheral lymphocyte counts induced by rIL-2 and numbers of cells isolatedfor culture (r = 0.79; P < 0.0001).

The data in Table 3 illustrate the large scale of the laboratorysupport required to carry out IL-2/LAK therapy. Mean totalnumber of cells (±SD)put into culture per patient was 14.3 ±7.4 x 10'°with a range of 5.1 to 44.2 x 10'°cells. This required,

4411

Research. on November 17, 2018. © 1988 American Association for Cancercancerres.aacrjournals.org Downloaded from

LABORATORY CORRELATES OF IL-2/LAK

5-1

? 4ox

72 3-

"35O I-

Leukopheresis Number

Fig. 3. Mean numbers of cells cultured for each day of leukapheresis. Barsindicate SEM.

Table 3 LAK cell culture data

PerpatientTotal

cells culturedMean ±SD

RangeCulture

litersMean ±SD

RangeLAK

cells infusedMean ±SDRange14.3

±7.5.1-44.295

±4641-2357.6

±3.32.0-16.24x

10'°x10'°x

10'°x 10'°

on average, 95 liters of cell culture per patient (range, 41 to235 liters). Total LAK cells harvested, then infused, per patientaveraged 7.6 ±3.3 x 10'°cells (range, 2.0 to 16.2 x 10'°cells).

This represents an average yield harvested from the cultures of53%.

To confirm that LAK effector cell activity was generated inthese cultures, aliquots of cells from culture Days 1 and 2 weretested in vitro for lysis of NK-resistant Daudi cells and/or freshtumor cells by a standard 5lCr release assay (2-4). Data from

67 of the 83 subjects were évaluablefor analysis of LAK activity.In 65 of these 67 cases (97%), fresh LAK cells caused >20%lysis of Daudi and/or fresh tumor cells at a 40:1 LAK celhtargetcell ratio. The mean percentage of cytotoxicity (±SD)underthese conditions was 72 ±22% against Daudi and 52 ±24%against fresh tumor cells. In general, there was a positive directcorrelation between cytotoxicity against Daudi and fresh tumorcell targets (r = 0.43; P = 0.001). Data from a representativepatient are shown in Fig. 4.

Relationships of LAK Numbers and Activity to Clinical Characteristics of Patients. Predictably, strong positive correlationswere identified by bivariate analysis between numbers of cellsestablished in culture and numbers of LAK cells subsequentlyharvested and infused into patients (Fig. 5a), and also betweentotal LAK cells infused and peak lymphocyte counts inducedby priming with rIL-2 (Fig. 5Z>).Table 4 gives results of univar-iate analyses of relationships of certain clinical characteristicsof patients with total numbers of LAK cells infused per patient.Although several trends are apparent, few significant correlations are established. Younger patients, males, and patientswith better performance status produced larger numbers ofLAK cells, but correlations did not achieve statistical significance. Similarly, patients who received 13 or more primingdoses of rIL-2 produced more LAK cells than those who received 12 or fewer doses (7.9 versus 6.6 x 10'°cells), but

differences between these groups also were not significant. Bycontrast, numbers of LAK cells infused did correlate with priorchemo- or radiation therapy (Table 4). Patients with no prior

K562 DaudiOvarianAdenocarcinoma

100n

80 40 IO 2.5

^Pre-LAK^<?—9—o-

80 40 IO 2.5

Effector ; Target Ratio

80 40 IO 2.5

Fig. 4. Ex vivo generation of LAK activity in a representative patient. Thepercentage of specific lysis calculated as described in "Materials and Methods" is

shown for patient cells incubated at 4 different effectortarget ratios with 3different "Cr-labeled targets: K562 and Daudi cultured cells; and fresh ovarian

adenocarcinoma cells. Results are given for unactivated peripheral mononuclearcells (pre-LAK) and for cells incubated for 3 days with rIL-2 (LAK).

20 n

'S

<n

<DS

15

10

5

n = 83R = .77p < .0001

•••

10 20 30

Cells Cultured

'—i—'40 50

20

15

10 n = 75R= 70p < .0001

500 1000 1500

Peak Lymphocyte Count2000

Fig. 5. a, bivariate plot of total LAK cells infused versus cells cultured, r =0.77 (P < 0.0001). b, bivariate plot of total LAK cells infused versus peaklymphocyte counts after rIL-2. r = 0.70 (P < 0.0001).

therapy generated and received an average of 8.3 x 10'°LAK

cells, whereas patients with prior therapy received only 6.4 xIO10LAK cells (P = 0.009). Patients with malignant melanomagenerated significantly more LAK cells (8.9 x 10'°)than pa-

4412

Research. on November 17, 2018. © 1988 American Association for Cancercancerres.aacrjournals.org Downloaded from

LABORATORY CORRELATES OF IL-2/LAK

Table 4 Relationships of clinical parameters with numbers ofLAK cells infused Table 5 Relationships of clinical parameters with activity ofLAK cells generatedex vivo

Numbers of patients for whom data with fresh LAK effectors were availablewere 66 with Daudi targets and 53 with fresh tumor cell targets.ParameterAge

<50(49)">50

(34)Sex

M(59)F

(24)Performance

status0(48)1(35)Disease

Renal cancer(32)Melanoma

(31)

Colon cancer(20)Doses

priming rIL-2si 2 doses(22)>12

doses(61)Prior

chemo- or X-ray therapyYes(32)No

(50)No.

ofLAKcells infused(cells x10-'°)8.0

±0.5*

6.9 ±0.58.0

±0.5

6.5 ±0.58.0

±0.57.0

±0.56.9

±0.58.9

±0.7

6.6 ±0.56.6

±0.7

7.9 ±0.46.4

±0.5

8.3 ±0.5P

=0.11P

=0.06/>=0.17P

=0.01P

=0.01P

=0.\3P=

0.009

" Numbers in parentheses, number of subjects.* Mean ±SEM.

tients with renal (6.9 x 10'°)or colon cancer (6.6 x 10'°).

However, as was the case for rIL-2-induced lymphocyte counts,this difference was accounted for by the younger age of melanoma patients and by the lower percentage of melanoma patients who had received prior therapy. Table 5 gives results ofunivariate analyses of relationships between clinical features ofthe patients and cytotoxic activities of their LAK cells generatedex vivo. No significant relationships were identified.

Multivariate analysis was used to identify factors which mightpredict yields or cytotoxic activities of LAK cells generated exvivo. Peak lymphocyte count was strongly predictive of LAKcell yields in this model (R2 = .49; P < .0001). Other variables

including age, sex, performance status, prior therapy, or actualnumbers of rIL-2 doses administered during priming were notpredictive of LAK cell yields. Although no variables in themultivariate model were identified which could predict LAKcell cytotoxicity against Daudi targets, the number of primingrIL-2 doses administered was moderately predictive of cytotoxicity against fresh tumor cells (P = 0.03).

Laboratory Correlates of Clinical Response to IL-2/LAK.Within the group of 83 évaluablepatients, there were 14 partialor complete responders. Table 6 compares 9 laboratory parameters in responding and nonresponding subjects assessed byunivariate analyses. No significant differences between responders and nonresponders were identified by base-line lymphocyte counts, peak lymphocyte counts induced by eitherpriming rIL-2 or therapeutic IL-2 plus LAK, cells cultured ortotal LAK cells infused per patient, doses of rIL-2 administered,or LAK cytotoxic activity against Daudi or fresh tumor targets.For the purposes of these analyses, cytotoxicity was quantitatedby indicating the percentage of specific lysis of target cells byfresh LAK effectors at a 40:1 effectontarget cell ratio.

Eleven independent variables were analyzed in a multivariate

Cytotoxic activity of LAK cells'

(% of specificlysis)ParameterAge

<=50>50Sex

MFPerformance

status01Disease

RenalcancerMelanoma

ColoncancerDoses

priming rIL-2<12doses>1

2dosesPrior

chemo- or X-raytherapy

YesNoDaudi

targets73

±4* (39)c/•=0.8171

±4(27)73

±3 (46)/>= 0.61

70 ±6(20)68

±4 (39)P = 0.06

78 ±4(27)70

±5 (23)P = 0.4\

75 ±4 (25)P = 0.97

70 ±6(18)76

±4 (10)P = 0.53

72 ±3(54)77

±4 (25)/>=0.18

69 ±3 (41)Fresh

tumortargets51

±5 (29)

53 ±5(24)53

±4(38)

47 ±6(15)52

±4 (38)

51±7(15)49

±5(18)

54 ±5(21)

53 ±6(14)46

±7 (9)

56 ±4(42)58

±6(18)

49 ±4 (35)P=0.71P

=0.34P

=0.84P

= 0.48

P =0.90P

=0.22p

=o.n

" Cytotoxic activity is quantitated as the percentage of specific lysis of Daudi

or tumor targets given by fresh LAK effectors at a 40:1 effectortarget ratio.* Mean ±SEM.c Numbers in parentheses, number of subjects.

Table 6 Laboratory correlates of clinical response to IL-2/LAK

ParameterTotal

LAK infused (cells xIO'10)LAK

activity (% of specificlysis)

Daudi targetsTumortargetsBase-line

lymphocyte count(cells/VOIL-2-induced

lymphocytepeak (a-llv',,1)IL-2/LAK-induced

lymphocyte peak(celiseli)Cells

cultured (cells xIO''0)IL-2

doses administered (no.of doses)

PrimingTherapeutic with LAKResponders7.7

±0.9°73

±657±61459

±3254612

±4875983

±263313.1

±1.513.4

±0.49.3 ±0.6Nonresponders7.5

±0.472

±350±41637

±1364926

+3156152

±48614.6

±0.813.3

±0.29.2 ±0.3Significance

level/>=0.85P

= 0.95P =0.45P

=0.60P

=0.67P

=0.92P

=0.43P

= 0.68P = 0.92

°Mean ±SEM.

model to attempt to identify factors which might be predictiveof patient responses to IL-2/LAK. These variables were age,sex, performance status, prior therapy, peak lymphocyte countsafter rIL-2 alone and after rIL-2 plus LAK, number of rIL-2doses given alone and with LAK cells, total LAK cells infused,and LAK cell activity against Daudi and fresh tumor targets.None was useful for predicting response in this model.

Laboratory Correlates of Clinical Toxicity Associated with II-

4413

Research. on November 17, 2018. © 1988 American Association for Cancercancerres.aacrjournals.org Downloaded from

LABORATORY CORRELATES OF IL-2/LAK

2/LAK. IL-2/LAK has been associated with a number of serioustoxicities (5-7), and the experience of the National CancerInstitute Extramural IL-2/LAK Working Group has confirmedthese observations.5 In the present study, 56 of the 83 évaluable

subjects experienced 109 episodes of Grade 3 or 4 toxicity.Major toxicities included hypotension (36 episodes), oliguria(21 episodes), gastrointestinal (13 episodes), CNS (12 episodes),pulmonary (12 episodes), cardiac (6 episodes), and other miscellaneous (9 episodes) manifestations. Table 7 compares 9laboratory parameters in subjects with severe (i.e., Grade 3 or4) or mild to moderate toxicities. No significant differencebetween the two groups was identified for any of these 9parameters. However, there was a tendency for severe toxicityto be associated with greater rIL-2-induced lymphocytosis,higher leukapheresis yields, and infusion of higher numbers ofLAK cells. It is noteworthy that numbers of priming or therapeutic rIL-2 doses administered did not correlate with toxicity.In large measure this was because rIL-2 doses were held fortoxicity. As a result, patients receiving less rIL-2 had equivalentor greater toxicity than those who continued to receive it.

As before, 11 independent variables were analyzed in a mul-tivariate model to attempt to identify factors which might bepredictive of severe toxicity during IL-2/LAK. Only one variable, i.e., cytotoxicity against Daudi targets, approached significance in this model (P = 0.06).

In additional analyses (not shown), each of the laboratoryparameters in Table 7 was examined for its potential relationship to the following specific toxicities: minimum hemoglobin;minimum leukocyte count; minimum platelet count; maximumbilirubin; maximum alkaline phosphatase; maximum serumglutamic oxaloacetic transaminase and serum glutamic pyruvictransaminase; maximum blood urea nitrogen; and creatinine.Although few correlations were identified by these analyses,those relationships which did emerge are of interest. Table 8shows correlations of 3 parameters, total LAK cells infused,leukapheresis yield, and peak rIL-2-induced lymphocyte count,with maximal recorded values of serum transaminases. A strongcorrelation was identified between transaminase levels and peakrIL-2-induced lymphocyte counts, and, to lesser extents, with

Table 7 Laboratory correlates ofIL-2/LAK-associated toxicity

Toxicity

Parameter Mils/moderate SevereSignificance

levelTotal LAK infused (cells X 6.9 ±0.5" 7.8 ±0.5 P = 0.25

LAK activity (% of specificlysis)

Daudi targetsTumor targets

71 ±6 73 ±3 P = 0.7649 ±5 53 ±4 P = 0.52

Base-line lymphocyte count 1801 + 326 1538 ±124 P = 0.36(cells/Ml)

IL-2-induced lymphocyte 4204 ±315 5132 + 354 P = 0.13peak (celiseli)

IL-2/LAK-induced lympho- 6243 ±1066 6096 ±604 P = 0.90cyte peak (cells'/ill

Cells cultured (cells x 10"10) 12.6 ±0.9 15.2 ±0.9 P = 0.08

11-2 doses administered (no.of doses)

Priming 13.1+0.3 13.4 ±0.2 /> = 0.51Therapeutic with LAK 9.4 ±0.5 9.2 ±0.4 P = 0.74«Mean ±SEM.

5 K. Margolin et al., manuscript in preparation.

leukapheresis yields and total LAK infused. By contrast, nocorrelation was identified between any of these parameters andthe peak levels of bilirubin or alkaline phosphatase (not shown).The other strong relationship to emerge from these analyseswas an association between cytolytic activity of LAK cellsagainst fresh tumor cells and oliguria (r = 0.47; P = 0.0002).

DISCUSSION

We report data derived from 83 patients treated with IL-2/LAK by the National Cancer Institute Extramural IL-2/LAKWorking Group. These patients were treated at 6 differentcenters according to a uniform protocol. Results afford anopportunity to examine closely the role of the laboratory component in IL-2/LAK immunotherapy. The data were analyzedto address two specific issues: (a) Can IL-2/LAK therapy beaccomplished outside of previously identified centers of expertise? (b) What are the relevant laboratory/clinical parametercorrelations?

The laboratory procedures required by IL-2/LAK therapy arecomplex, cumbersome, and technically demanding. In additionthey are labor and material intensive. The average patient inthis study generated 95 liters of cell cultures (range, 41 to 235liters) and received 7.6 x 10'°LAK cells reinfused. As judgedby the percentage of specific lysis >20% in "Cr release cytotox

icity assays, LAK cells with effective in vitro tumoricidal activityagainst cultured Daudi and fresh tumor cells were produced in97% of évaluablecases. These data demonstrate that largenumbers of LAK effector cells with in vitro tumoricidal activityin sufficient quantities for therapeutic administration were produced by all 6 extramural centers. Despite the "industrial scale"

of these laboratory requirements, it is clear that ex vivo LAKcell generation can be accomplished routinely and effectively atdifferent sites. Although it is unlikely that IL-2/LAK therapywill or should be performed routinely in every hospital, thetechnology is manageable in those institutions with establishedapheresis units, large scale cell culture facilities, and the requisite capabilities for intensive clinical support of IL-2/LAK

patients.It was of interest to examine clinical parameters that might

influence numbers or cytotoxic activities of LAK cells generatedex vivo. The only clinical parameter which correlated significantly with numbers of LAK cells infused was a history of priorchemo- or radiation therapy (Table 4). Patients with no priortherapy generated more LAK cells than those with prior therapy. Although other differences did not achieve statistical significance, the data suggest that age <50, male sex, performancestatus = 0, and >12 priming doses of rIL-2 may influencepositively the number of LAK cells generated. In a multivariatemodel the only factor predictive of LAK cell yields was peaklymphocyte count after infusion of rIL-2. By contrast, no clinical parameters appeared to influence in vitro cytotoxic activityof LAK cells when tested by univariate analysis (Table 5). And,in a multivariate model, only the number of priming rIL-2doses administered had any predictive value for LAK cytotoxicity, and this was of modest significance.

This trial confirmed the effect of rIL-2 administration onperipheral lymphocyte counts previously observed by others (7,10, II). Lymphopenia and rebound lymphocytosis were observed in response to both priming and therapeutic rIL-2 administration (Fig. 2). Clinical correlations with lymphocytosisinduced by priming doses of rIL-2 are listed in Table 2. Severalclinical factors including age <50, no prior therapy, and numberof doses of priming rIL-2 correlated positively with extent of

4414

Research. on November 17, 2018. © 1988 American Association for Cancercancerres.aacrjournals.org Downloaded from

LABORATORY CORRELATES OF IL-2/LAK

rebound lymphocytosis in our patients. West et al. also reportedthat rebound lymphocytosis correlated directly with dose ofpriming rIL-2 (7).

Leukapheresis yields and, ultimately, numbers of LAK cellsgenerated on each of the 5 days of apheresis reflected theperipheral lymphocyte counts and were positively correlatedwith rebound lymphocytosis. Highest LAK cell yields wereachieved on Day 1 of leukapheresis and declined steadily thereafter. Average cell yield on the last day of leukapheresis wasless than half the yield on Day 1, representing approximately13% of the total number of cells put into culture. These datasuggest that more intensive apheresis earlier in the course ofrIL-2-induced rebound lymphocytosis might be more productive. Subsequent protocols piloted by the Extramural WorkingGroup have been designed to take advantage of these observations. These protocols will be the subject of subsequent reportsfrom the Group.

In the report of West et al., responding patients had significantly higher base-line and rebound lymphocyte counts thannonresponders (7). In their study of 40 patients treated withcontinuous-infusion rIL-2, 11 of 15 subjects who achieved rlL-2-induced rebound lymphocyte counts of 6000 per p.\or greaterhad meaningful tumor reductions, whereas tumor reductionswere seen in only 2 of 25 subjects who achieved lower reboundcounts. In our study we observed no differences in base-linelymphocyte counts, rIL-2-induced lymphocyte peaks, or rIL-2/LAK-induced lymphocyte peaks between responding and non-responding patients (Table 6). Nineteen of the 83 subjectsachieved lymphocyte counts >6000 per ¡Aafter priming rIL-2.Of these 19 patients, 4 were responders (21.1 %). This comparesto a lower response rate of 10 of 64 or 15.6% for patients withpeak lymphocyte counts <6000 per ^1 which is not significantlydifferent. However, it is noteworthy that two of the four patientswho achieved complete responses were in the group of subjectswith rebound lymphocytosis >f>000 per n\. Although these datado not confirm unequivocally the reported observations of Westet al. (7), the trend indicated by our results is consistent withtheir observation that subjects achieving the highest reboundlymphocyte counts after rIL-2 are those more likely to havetumor responses. The relationship between rebound lymphocytosis and tumor responses might reflect individual sensitivityto biological effects of rIL-2. This possibility is supported bythe observation of a strong positive correlation between theheights of the first and second rebound lymphocyte peaks inthis group of patients. Also, the data suggest that, for optimaltherapeutic efficacy, future schedules for administering thepriming doses of rIL-2 should be devised with the goal ofmaximizing rebound lymphocyte counts.

Table 6 shows relationships between key laboratory parameters and patient responses to IL-2/LAK therapy. No significantcorrelations were identified. Table 7 shows relationships between key laboratory parameters and IL-2/LAK-associated tox-icities. Neither numbers nor in vitro cytotoxic activities of LAKcells administered differed between patients with mild/moder-

Table 8 Laboratory correlates of IL-2/LAK-associated transaminasemia

Relationships between variables were analyzed using Pearson correlation coefficients.

SCOT"

SGPTPeak

lymphocytecount after rIL-2(r)0.63

(/>< 0.0001)0.73 (P< 0.000 1)Leukapheresis

yield(r)0.34

(P = 0.004)0.37 (P = 0.003)Total

LAKinfused

(r)0.30

(P= 0.01)0.31 (/>=0.01)

" SCOT, serum glutamic oxaloacetic transaminase; SGPT, serum glutamic

pyruvic transaminase.

ate or severe toxicities. In addition, multivariate analyses failedto identify any clinical or laboratory variable that was predictiveof patient responses to IL-2/LAK. Neither numbers of LAKinfused nor the in vitro cytotoxic activity of infused LAKdiffered between responders and nonresponders. Similar to theresults in this study, West et al. reported that, even thoughresponding patients received more LAK cells and more totalcytolytic units than nonresponding patients, neither of thosedifferences was significant (7).

Interpretation of these observations is a complex issue sincemultiple additional factors may be important. These include,among others, the variable heterogeneity of the rIL-2-activatedcells infused, in vivo LAK activity, the tumor burden, and theaccessibility of the tumor to the infused cells. In fact, the datain both Tables 6 and 7 showing no relationships betweennumbers or activities of LAK cells administered to patients orrIL-2 doses administered and tumor responses or clinical toxicities dramatize the need to identify other parameters thatmight be useful for monitoring IL-2/LAK therapy. Because ofthese considerations, the important question about the need forex vivo LAK cell generation in treatment of human cancers withrIL-2 will be answered definitively only by a carefully controlledprospective trial.

The strong relationship between peak lymphocyte countsachieved after priming rIL-2 and elevated serum transaminasein these patients is of interest (Table 8). Such transaminaseelevations may serve as a sensitive indication of hepatic infiltration by IL-2 activated lymphocytes. The extent to which suchactivated cells are hepatotoxic is a question raised by these data.Other tests of hepatic integrity (alkaline phosphatase, totalbilirubin), although frequently abnormal, did not correlate withlymphocytosis in a similar manner. Furthermore, in all casesabnormalities of liver function tests returned to base-line valuesafter completion of therapy, indicating that the insult to theliver was transient and reversible.

Likewise the relationship between LAK cell activity againsttumor targets and oliguria is of interest. Oliguria in thesepatients predominantly was a reflection of decreased intravas-cular volume secondary to the increased capillary permeabilitywhich has been strongly associated with IL-2/LAK therapy (1,5-7, 12). The mechanism by which IL-2/LAK induces a vascular leak syndrome is unknown, but data indicate that thiseffect may be mediated directly or indirectly by host lymphoidelements (13). In addition, other data now indicate that LAKcells bind to and lyse normal human vascular and cornealendothelial cells in vitro (14, 15). The association we haveobserved between LAK antitumor cell activity generated ex vivoand oliguria, in fact, may be an indication of such lysis ofendothelial cells in our patients by highly active LAK cells.

ACKNOWLEDGMENTS

We thank Dr. Gary M. Clark for providing invaluable assistancewith data analysis. Secretarial support was provided by J. Skinner andK. Kaufman.

REFERENCES

1. Rosenberg, S. A. Adoptive immunotherapy of cancer using lymphocyte-activated killer cells and recombinant interleukin-2. In: V. T. DeVita, S.Hellman, and S. A. Rosenberg (oils.). Important Advances in Oncology, pp.55-91. Philadelphia: J. B. Lippincott, 1986.

2. Lotze, M. T., Grimm, E. A., Mazumder, A., Strausser, J. L., and Rosenberg,S. A. Lysis of fresh and cultured autologous tumor by human lymphocytescultured in T-cell growth factor. Cancer Res., 41: 4420-4425, 1981.

3. Grimm, E. A., Mazumder, A., Zhang, H. Z., and Rosenberg. S. A. Lympho-

4415

Research. on November 17, 2018. © 1988 American Association for Cancercancerres.aacrjournals.org Downloaded from

LABORATORY CORRELATES OF IL-2/LAK

kine-activated killer cell phenomenon: lysis of natural killer-resistant freshsolid tumor cells by interleukin 2-activated autologous human peripheralblood lymphocytes. J. Exp. Med.. 755: 1823-1841, 1982.

4. Rosenstein, M., Yron. l., Kaufmann, Y., and Rosenberg, S. A. Lymphokine-activated killer cells: lysis of fresh syngeneic natural killer-resistant murinetumor cells by lymphocytes cultured in interleukin 2. Cancer Res., 44: 1946-1953, 1984. '

5. Rosenberg, S. A., Lotze, M. T., Muul, L. M., et al. Observations on thesystemic administration of autologous lymphocyte-activated killer cells andrecombinant interleukin-2 to patients with metastatic cancer. N. Engl. J.Med.. 313: 1485-1492, 1985.

6. Rosenberg. S. A., Lotze, M. T., Muul, L. M., et al. A progress report on thetreatment of 157 patients with advanced cancer using lymphocyte-activatedkiller cells and ¡nterleukin-2 or high-dose interleukin-2 alone. N. Engl. J.Med., 316: 889-897, 1987.

7. West, W. H., Tauer, K. W., Yannelli, J. R. Marshall, G. D., Orr, D. W.,I Immuni. G. B., and Oldham. R. K. Constant infusion recombinant interleukin-2 in adoptive immunotherapy of advanced cancer. N. Engl. J. Med., 316:898-905, 1987.

8. Muul, L. M., Director, E. P., Hyatt, C. L., and Rosenberg, S. A. Large scaleproduction of human lymphocyte activated killer cells for use in adoptiveimmunotherapy. J. Inumino! Methods, 88: 265-275, 1986.

9. Dixon, W. J. BMDP Statistical Software. Berkeley, CA: University of California Press, 1981.

10. Lotze. M.T., Matory, Y. L., Ettinghausen, S. E., et al. In vivoadministrationof purified interleukin-2. II. Half life, immunologie effects, and expansion ofperipheral lymphoid cells in vivo with recombinant IL-2. J. Immunol., 135:2865-2875, 1985.

11. Atkins, M. B., Gould, J. A., Allegretta, M., Li, J. J., Dempsey, R. A.,Rudders, R. A., Parkinson, D. R., Reichlin, S., and Mier, J. W. Phase Ievaluation of recombinant interleukin-2 in patients with advanced malignantdisease. J. Clin. Oncol., 4: 1380-1391, 1986.

12. Belldegrun, A., Webb, D. E., Austin, H. A., Ill, Steinberg, S. M., White, D.E., Linehan, W. M., and Rosenberg, S. A. Effects of interleukin-2 on renalfunction in patients receiving immunotherapy for advanced cancer. Ann.Intern. Med., 106: 817-822, 1987.

13. Rosenstein, M., Ettinghausen, J. E., and Rosenberg, S. A. Extravasation ofintravascular fluid mediated by the systemic administration of recombinantinterleukin 2. J. Immunol., 137: 1735-1742, 1986.

14. Damle, N. K., Doyle, L. V., Bender, J. R., and Bradley, E. C. Interleukin-2-activated human lymphocytes exhibit enhanced adhesion to normal vascularendothelial cells and cause their lysis. J. Immunol., 138: 1779-1785, 1987.

15. Kotasek, D., Ochoa, A. C., Vercellotti, G. M., Bach, F. H., and Jacob, H. S.I AK cell-mediated endothelial injury: a mechanism for capillary leak syndrome in patients treated with I .AK cells and IL-2. Clin. Res., .Õ5:660A,1987.

4416

Research. on November 17, 2018. © 1988 American Association for Cancercancerres.aacrjournals.org Downloaded from

1988;48:4409-4416. Cancer Res   D. H. Boldt, B. J. Mills, B. T. Gemlo, et al.   Cells in HumansRecombinant Interleukin-2 and Lymphokine-activated Killer Laboratory Correlates of Adoptive Immunotherapy with

  Updated version

  http://cancerres.aacrjournals.org/content/48/15/4409

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/48/15/4409To request permission to re-use all or part of this article, use this link

Research. on November 17, 2018. © 1988 American Association for Cancercancerres.aacrjournals.org Downloaded from