The Effects of Gamma Interferon on the Natural Killer and Tumor Cells of Children With Neuroblastoma

A Preliminary Report

AUDREY EVANS, MD,*it ELLIOTT MAIN, MD,* KAREN ZIER, PHD,’ NAOHIKO IKEGAKI, PHD,f MARGARET TARTAGLIONE,t ROGER KENNETT, PHD,t AND LOIS LAMPSON, PHDS

Human neuroblastoma cells lack HLA-A,-B,-C molecules which can be induced in vitro by gamma in- terferon (yIFN). To test the hypothesis that the same induction would occur in vivo leading to tumor regression, a Phase I study was initiated. Seven patients with neuroblastoma were entered on a Phase I study of recombinant yIFN in children. Three received 0.05 mg/m2 intravenously (IV) three times a week, three received 0.1 mg/m2 for 4 weeks, and one patient withdrew from study before receiving adequate treatment for evaluation. No significant clinical response was seen. The side effects were fever and chills, and no serious toxicity occurred. Natural killer (NK) and lymphocyte activated killer (LAK) precursor activity of peripheral blood mononuclear cells was determined before and during treatment, and expression of HLA-A,B,C molecules was looked for on the tumor cells in the bone marrow of five patients. The NK activity initially low, reached control levels in six patients, but the increase was transient. The LAK precursor activity remained normal. Expression of HLA-A,B,C, initially absent, was induced on the neu- roblastoma cells in four of six patients.

Cancer 64: 1383-1387, 1989.

EUROBLASTOMA, a tumor of the sympathetic ner- N vous system seen in early childhood, has various degrees of malignancy. The spectrum ranges from spon- taneous regression of bulk disease to relentless progression despite aggressive therapy. The immune response, either effective or ineffective, has been implicated as a mediator in the progression or regression of this tumor. In the 1960s, the Hellstrom et al. showed that there were cytotoxic lym- phocytes to neuroblastoma (NBL) and also demonstrated serum blocking factors.’’2 In a study of patients at this institution, Hann and associates have demonstrated that those with elevated serum femtin have a factor (presum- ably fenitin) that inhibits the ability of T-cells to form rosettes with sheep erythrocyte^.^ Several investigators have associated an increased peripheral lymphocyte count

From the Children’s Cancer Research Center, The Children’s Hospital of Philadelphia, Philadelphia; Departments of *Pediatrics,and tHuman Genetics, the University of Pennsylvania, Philadelphia; the $Department of Obstetrics and Gynecology, University of California, San Francisco; and the §Department of Neurology, Harvard Medical School, Boston.

Supported by funds from the Seligson Foundation and grant no. CA- 14489 from the National Cancer Institute.

The authors thank Genentech (San Francisco, CA) for supplying the yinterferon.

Address for reprints: Audrey E. Evans, MD, Division of Pediatric Oncology, The Children’s Hospital of Philadelphia, 3400 Civic Center Boulevard, Philadelphia, PA I9 104.

Accepted for publication February 20, 1989.

with a better outcome, and lymphocytic infiltration of the tumor has been associated with a favorable progno~is .~ ,~ Data from a Childrens Cancer Study Group protocol for patients with local and regional NBL showed a correlation between an absolute lymphocyte count above 3000 mm2 at diagnosis and a better outcome independent of age and stage.6

Recent work draws attention to two cellular immune mechanisms that could be relevant for NBL. The first is natural killer (NK)-mediated cytotoxicity. Alvarado and co-workers reported that NK activity in children with malignant solid tumors including some with NBL, is re- duced compared to normal control^.^ Previous work by Main and colleagues has shown that NBL cell lines in vitro are sensitive to NK c e k 8 The second cellular mech- anism of potential interest is T-cell-mediated cytotoxicity. Lampson and colleagues have shown that NBL cells lack Class I antigens, and that they are insensitive to cell-me- diated cytotoxicity by a T-cell line directed against Class I antigens shared by the original sensitizing cell line and the NBL target.’ They have also shown that in v i m , gamma interferon (yIFN) induces Class I mRNA and surface antigens on NBL cell lines.lO~ll

Based on the work of Lampson et a1.’-“ showing in- duction of Class I antigenic expression in NBL cells, and of Main et a1.’ showing depressed NK function in patients

1383

1384 CANCER October 1 1989 Vol. 64

with NBL, a Phase 1/11 study of yIFN in the treatment of children with NBL resistant to conventional treatment was initiated. Here we describe the effects of the initial low dose on both HLA-A,B,C expression of the cancer cells and the immunologic function of the host as assessed by assays of circulating NK and LAK cells and of LAK precursors. This is the first report on the effects of yIFN in children with NBL, and the first demonstration of the in vivo induction by yIFN of Class I expression on a tumor not normally expressing those antigens.

Methods

Patient Population

The patients with NBL eligible for study were those resistant to conventional treatment, had a life expectation of 6 weeks, had no significant organ impairment, and were entered on study by the parents signing the study consent form.

The initial dose of yIFN was 0.05 mg/m2 given intra- venously (IV), Monday, Wednesday, and Friday, for 4 weeks. This is 10% ofthe maximum tolerated dose (MTD) given in this fashion to adults. Dose escalations are planned at 0.10, 0.25, 0.50, and 0.75 mg/m2. The yIFN was supplied by Genentech (San Francisco, CA) in a preparation of 2 x 10' U/mg protein.

Peripheral blood was obtained weekly for studies of NK and LAK function, and in patients with tumor infil- trating the bone marrow, aspirations were done at intervals of 10 days during treatment.

Analysis of Natural Killer and Lymphocyte Activated Killer (LAK) Function

All assays were done using freshly isolated mononuclear cells as previously described.8 In brief, spontaneous NK activity was tested by mixing serial dilutions of mono- nuclear cells isolated from peripheral blood together with a constant number of chromium 5 1 ("Cr)-labeled autol- ogous lymphocytes, K562 cells, or Daudi cells. Sponta- neous NK cells are able to lyse K562 cells well, but they lyse Daudi cells either poorly or not at all. The LAK cells lyse both targets with approximately equal ability. After incubation at 37°C for 4 hours, the cells were pelleted and supernatant samples counted for release of "Cr. As- sessment of precursors able to generate LAK activity in vitro was also made. The procedure was similar to that for NK activity with the exception that effector cells were generated by culturing lymphocytes for 6 days in chro- matographically purified/recombinant interleukin-2 (IL- 2) before being mixed with target cells. Results were ex- pressed as either percent cytotoxicity or number of lytic units/ lo6 cells.''

Tumor Cells in the Marrow

Tumor cells in the marrow are studied by means of immunofluorescence as previously de~cribed.'~ The mar- row was separated by ficoll-hypaque density centrifuga- tion, and the interface layer harvested and washed with phosphate-buffered saline (PBS). The cells were incubated for 30 minutes with two monoclonal antibodies, PI 153/ 3, which recognizes NBL, and KE 2, which identifies HLA-A,B,C. Unconcentrated supernatant antibodies were used 25p1 PI 153/3 and 5Opl KE2. After incubation the cells were washed two times in a balanced sodium-po- tassium buffer with 0.1 % gelatin, and were incubated for a further 30 minutes with two goat anti-mouse antibodies, one rhodamine-labeled anti-IgM, the other fluorescein- labeled anti-IgG, thus differentiating the first two anti- bodies. After the 30-minute incubation and washing, the cells were placed on glass slides without fixing and eval- uated by fluorescence microscopic study. The NBL cells and some reacting pre-B-lymphocytes were visible on the rhodamine wavelength (Leitz Filter System N), and the normal marrow elements were visible on the fluorescein wavelength (Leitz Filter System L2). The PI 153/3 "false- positive" lymphocytes were double-labeled since they also have HLA-A,B,C antigens. The NBL cells are not visible on the fluorescein wave length. Early expression of HLA- A,B,C on tumor cells was seen as a green speckling of the rhodamine-positive cells visible under the fluorescein wave length. Background light helps to confirm that these cells were indeed tumor cells in clumps.

Western Blot Analysis of Tumor Cells

The tumor cells in one patient grew in vitro and a cell line was established. To study the effect of yIFN in vitro, these cells were grown in the presence of increasing amounts of yIFN and were harvested after 6 days. They were viable and expression of HLA-A,B,C surface antigen was confirmed by immunofluorescence microscopic study. Western blot analysis was done on the cells that were exposed to 50 U of yIFN, and the results compared to the cells that were not so exposed. The cell extract was prepared by lysis of the cells in PBS containing 0.5% NP- 40 at 100 mg wet cells per ml and the lysate was centri- fuged in an Eppendorf centrifuge for 1 minute. A clear supernatant free of the cell nuclei was collected as cyto- plasmic/membrane fraction. Thirty microliters of the original and two-fold diluted samples of this fraction were subjected to sodium dodecylsulfate/polyacrylamide gel (SDS/PAGE). The protein separated on the gel was then transferred to a nitrocellulose paper by electrophoretic manner.I4 After blocking the paper with 2% bovine serum albumin (BSA), 2% Ficol and 2% polyvinylpyrrolidone in 50 mmol/l Tris HCl, pH 7.6, an anti-HLA-A,B,C monoclonal antibody 020 1-2027 (purchased from Cooper

No. 7 yIm FOR CHILDREN WITH NEUROBLASTOMA * Evans el a[. 1385

TABLE I . LvmDhocvte Counts and ResDonse to rIFN.

Dose Patient (-//m2)

Absolute lymphocyte count X lo3 NK activity (% of control)

Owk 1 wk 2wk 3wk 4 w k Owk 1 wk 2wk

1 0.05 x 12 2 0.05 x 11 3 0.05 x 12 4 0.1 x 3 5 0.1 x 12 6 0.1 x 12 7 0.1 x I 2

1.638 0.910 1.898 0.852 0.315 0.286 1.127 0.552 0.264 - 0.520 0.306 0.356 0.032 0.420 0.286 1.040 - - - 0.170 0.250 0.220 0.360 0.830 0.316 - 0.273 0.576 0.483 0.496 1.152 1.476 1.326 0.882

33 87 118 44 29 - 0 14 0 0 - -

23 14 7 48 180 100 40 18 -

3 wk 4 wk Tumor HLA expression

370 22 None - +d 21

78 10 +d21 and28 - - Inadequate trial 74 14 10%-15% cells+ on d 28 - 57 No tumor to test

4 17 10%-15% cells+ on d 28

-

NK: natural killer cell; HLA: human leukocyte antigen.

Biomedical, Inc., Malvern, PA) was applied. To visualize the specific reaction of the monoclonal antibody, the blot was incubated sequentially with biotinylated goat anti-mouse antibodies, peroxidase-conjugated Streptavi- din, and substrate/chromogen mixture (H202/4-chloro- 1 naphtol) from Sigma (St. Louis).

Results Seven patients have been entered on study, and one is

excluded from this report as he received only three doses and no further studies were done before he was withdrawn from treatment; five of six completed the course of 12 injections with yIFN at 0.05 mg/m2 or 0.1 mg/m2. Patient 2 died after 11 treatments. All but one had widespread Stage IV disease with involvement of liver, bone, and bone marrow. In Patient 1 there was a moderate decrease in the size of a supraclavicular mass of nodes which regressed from 6 to 3 cm during treatment, and at the end of treat-

ment, were more easily identified as separate involved lymph nodes.

Other areas of disease failed to show a response. Four of six patients experienced fever and chills with maximum temperatures of 39.5"C and 40°C. One patient developed an increase in bilirubin and liver enzymes, which was thought most likely the result of disease progression as seen by ultrasound rather than an effect of yIFN.

All patients were anemic and received blood transfu- sions, and all but one were lymphopenic before initiation of therapy (Table 1). In two of the five lymphopenic pa- tients, there was an increase in the total lymphocyte count during treatment from 286 to 1 127 and 496 to 1476 mm3, respectively, but the increase was not sustained. The NK activity is given as a percentage of the normal controls run on the day of the experiment. It was reduced or absent initially in all patients. It can be seen in Table 1 that there was an increase to near normal levels in all but Patient

FIG. 1. NK activity in peripheral blood cells of Patient 4. The results are compared to two normal donors who did not receive -/IF". Increased activity is seen at week 3 when the level equalled that of one of the nor- mal donors.

1386 CANCER October 1 1989 Vol. 64

7, and in Patients 1 and 6 the activity was greater than the control. The time of the increase was not uniform, varying from week 1 to week 3, and did not correlate with an increase in the absolute lymphocyte count. The NK results obtained using cells from Patient 5 are illustrated

FIG. 3. Western blot analysis of NBL cells from Patient 3 exposed in v i m for 6 days to 50 U of $FN. Lane 1 represents cells not exposed to yIFN and Lanes 2 through 8 represent serial dilutions of the exposed cells showing a 32-fold increase in the amount of HLA expressed.

FIG. 2. LAK activity in the same patient and normal donors showing similar responses throughout treat- ment with yIFN.

in Figure 1. All experiments were carried out using fresh blood and the same normal controls were used during the entire treatment period. No patient had spontaneous (cir- culating) LAK activity before or during therapy. In con- trast, the ability to generate LAK activity ( ie . , LAK pre- cursors) was normal before treatment and remained so during the 4-week course of yIFN (Fig. 2).

Five patients had tumor in the marrow. In the first patient, HLA-A,B,C antigen was measured on fixed bone marrow smears by peroxidase-antiperoxidase (PAP assay) at the beginning and end of treatment. No HLA-A,B,C was detected on the tumor cells in this case. In the other four patients, the marrow was studied by means of im- munofluorescence. At 10 days, after four injections of yIFN, the NBL cells in two patients were faintly reactive with KE2. This was detected by dots of green fluorescence on the PI 153/3-positive cells. By 28 days stronger expres- sion was detected. In the other two patients, at 28 days, 10% to 15% of tumor cells showed minimal reaction with the antibody to HLA-A,B,C antigen. To insure that the positive tumor cells were synthesizing HLA-A,B,C mol- ecules, and had not passively adsorbed circulating antigen, the immunofluorescence assay was repeated on cells that had been cultured in vitro (without additional interferon) for 3 days. The cells were fixed with 3.4% formaldehyde and studied with KE2 and a fluorescein-labeled second antibody. As before, fluorescence was detected in the membrane, but there was also faint staining in the cyto- plasm.

The tumor cells obtained from the marrow of Patient 3 at the time of study entry, were grown in culture and a

No. 7 yIFN FOR CHILDREN WITH NEUROBLASTOMA * Evans et a/. 1387

cell line was established. Since it is difficult to physically separate pure tumor cells from the marrow, this cell line was used to explore further the effect of yIFN. Western blot analysis of the cells exposed in vitro to yIFN showed at least a 32 times increase in the amount of HLA ex- pressed compared to the same cells grown under standard conditions (Fig. 3 ) . This confirms previous work by Lampson et af. with long-term cell lines. ''

Discussion

This study reports our observations on the effects of yIFN on NK and LAK activity. Even though the dose levels of 0.05 and 0.1 mg/m2 were considered subthera- peutic, increase in NK activity has been seen after daily intramuscular (IM) doses of 0.01 and 0.1 mg/m2 in pa- tients with melan~ma. '~ The increase in NK activity seen in this study was transient and the time from initiation of treatment to response varied from 1 to 3 weeks. There are two possible explanations for the transient response: all patients had had extensive prior therapy at time of entry on study, and all had initially low or undetectable activity. The patients in the study by Maluish et af. all had minimal disease and no prior chemotherapy." Sec- ondly, the intermittent 3 day a week IV route of admin- istration may be a factor in induction and maintenance of response.

It was also possible to show in the target cells that the HLA-A,B,C was expressed strongly in two patients and weakly in an additional two patients. The observed in- crease in Class I expression on NBL is a further basis for expanding our studies designed to measure T-cell func- tion, since Class I-positive tumor cells might be candidate targets for T-effector cells in vivo. This was not observed in vitro by Maine et aL8 but has more recently been dem- onstrated by Zier et a/. l 6 Furthermore, the demonstration of normal LAK cell precursor activity after in vitro stim- ulation with IL-2 suggests a rationale for an IL-2 trial. These studies demonstrate for the first time that yIFN administered in vivo can affect both NBL cell surface properties and the patient's immune effector mechanisms.

They form the basis for interpretation and further analysis of clinical effects that may be seen as the study is extended to more therapeutic doses of yIFN.

REFERENCES

1. Hellstrom I, Hellstrom KE, Pierce GE et al. Demonstration of cell bound and humoral immunity against neuroblastoma cells. Proc Nut1 Acad Sci USA 1968; 60:1231-1238.

2. Hellstrom I, Hellstrom KE, Bill AH et al. Studies on cellular im- munity to human neuroblastoma cells. Int J Cancer 1970; 6: 172-1 88.

3. Hann HW-L, Evans AE, Cohen U et al. Biological differences between neuroblastoma Stages IV-s and IV. N Engl J Med 198 1 ; 304

4. Bill AH. The implications of immune reactions to neuroblastoma. Surgery 1962; 66:415-418.

5 . Bill AH, Morgan A. Evidence for immune reactions to neuroblas- toma and future possibilities for investigations. J Pediatr Surg 1970; 5:

6. Evans AE, Albo V, DAngio GJ et a/. Factors influencing survival of children with non-metastatic neuroblastoma. Cancer 1976; 38:66 I - 666.

7. Alvarado C, Findley H, McKolanis J et al. Natural killer cells in children with malignant solid tumors. Proc Am Soc Clin Oncol 1987; 6: 218.

8. Main EK, Lampson LA, Hart MK et al. Human neuroblastoma cell lines are susceptible to lysis by natural killer cells but not by cytotoxic T-lymphocytes. J Immunol 1985; 185:242-246.

9. Lampson LA, Fisher CA, Whelan JP. Striking paucity of HLA- A,B & C and (32 microglobulin on human neuroblastoma cell line. J Immunoll983; 130:247 1-2478.

10. Lampson LA, George DL. IFN-mediated induction of Class I MHC products in human neuronal cell lines: Analysis of HLA and p2- mRNA and HLA-A, and HLA-B protein and polymorphic specificities. J Interferon Res 1986; 6:257-265.

1 I . Lampson LA, Fisher CA. Weak HLA and p2 micro globulin expression of neuronal cell line can be modulated by interferon. Proc Natl Acad Sci USA 1984; 81:6476-6480.

12. Pross HF, Baines MG, Rubin P et al. Spontaneous human lym- phocyte-mediated cytotoxicity against tumor target cells: IX. Quantitation ofNK activity. J Clin Immunol 1981; 151-61.

13. Evans AE, Griffin GC, Tartaglione M et al. A method ofdetecting neuroblastoma in human bone marrow by means of two monoclonal antibodies P1153/3 and KE2. Hybridoma 1985; 4:289-296.

14. Towbin H, Staehelin T, and Gopdon J. Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: Procedure and some applications. Proc Natl Acad Sci USA 1979; 76:4350-4354.

15. Maluish AE, Urba WJ, Longo DL et al. The determination of an immunologically active dose of interferon-gamma in patients with mel- anoma. J Clin Oncol 1988; 6:434-445.

16. Zier KS, Pierson GR, Brown V. Susceptibility of human neuro- blastoma cell lines to CTL-mediated lysis (in preparation).

425-429.

11 1-1 16.