Managing clinical research data: software tools for hypothesis exploration.

Data representation, data file specification, and the communication of data between software systems are playing increasingly important roles in clinical data management. This paper describes the concept of a self-documenting file that contains annotations or comments that aid visual inspection of the data file. We describe access of data from annotated files and illustrate data analysis with a few examples derived from the UNIX operating environment. Use of annotated files provides the investigator with both a useful representation of the primary data and a repository of comments that describe some of the context surrounding data capture.

The treatment of malignant diseases by passive administration of antibodies (Wright & Bernstein 1980) has been pursued as a promising therapy, although the life span after antiserum administration has not so far been improved. The failure of conventional serotherapy has been mainly attributed to the low specificity of the antisera and the poor immunogenicity of tumour-associated antigens (TAA). The new era in serology since hybridoma technology (Kohler & Milstein 1975) was devised might be exploited for new modalities of cancer immunotherapy.
In a previous report from this laboratory (Testorelli et al., 1982), the production and the characterization of a MAb reacting specifically with the cells of the L1210 murine leukaemia (L.1) was described. The L.1 MAb displayed in vitro very high and specific complement-dependent cytotoxicity against L1210 cells. On the other hand, it did not react with normal adult and foetal murine tissues, with cells from several independent haplotypes or with a number of chemicaland virus-induced experimental tumours.
In the present report, the therapeutic effects of the administration of L.1 alone or combined with chemotherapeutic agents on the survival rates of leukaemic mice are presented. Possible reasons for the ineffectiveness of serotherapy with L.1 were also investigated and are discussed.
Hybrid cell lines and production of ascitic fluid The production of the hybridoma cell line secreting the monoclonal antibody L. 1 (IgM) to a surface antigen of L1210 cells has been described previously (Testorelli et al., 1982). The hybridoma was maintained in culture in RPMI 1640 medium (Eurobio, Paris), supplemented with 10% heatinactivated foetal calf serum (FCS, Seromed GmBg, Miinchen, Germany) and lOO1 I mlgentamycin (Seromed). The supernatant from the cultures was stored at -70°C. For production of ascitic fluid, Balb/c or CD2F, mice, preprimed with pristane (0.5ml,i.p.), were injected i.p. with 106 hybrid cells. Ascitic fluid was collected after 15-20 days, centrifuged to remove the cells and frozen at -70"C until use. The cytotoxic titers of both supernatant and ascitic fluid were assessed by a complement-dependent cytotoxicity assay (see below) using Cr-labelled L 1210 cells as target. Titres are expressed as the highest dilution that killed 50% of leukaemic cells. When not otherwise specified, the cytotoxic titre was 1:64 for the supernatant from the hybridoma culture and 10-6 for ascites. Complement-dependent cytotoxicity assay (CDC) 51Cr-release assay was performed as described elsewhere (Testorelli et al., 1982). Briefly, 25,l of target cells labelled with [51Cr] sodium chromate were incubated with 25 jl of the L.1 (P3X63/Ag8 supernatant was used as control) for 45 min at 37°C. Rabbit complement diluted 1:5 was then added (25 ,l) and the cells were incubated for 30min at 37°C in a humidified 5% CO2 atmosphere. The percentage of Cr release into the supernatants of triplicate samples was calculated according to Ozato et al. (Ozato et al., 1980).

Indirect membrane immunofluorescence (IMF)
The L1210 cells were tested after extensive washing for their viability (cells <95% viable were excluded) and incubated with 20,l of supernatant from the hybridoma culture for 40 min at 4 C. The cells were washed twice and subsequently incubated with 0.15 of a 1:10 dilution of fluorescein isothiocyanate-conjugated (FITC) rabbit anti-mouse IgG (Cappel Lab., Cochranville, PA) for 30min at 4°C. The cells were washed 3 x and analyzed under a microscope with epifluorescence optics and in the fluorescence-activated cell sorter (FACS II, Becton Dickinson, Mountain View, Calif.). Serotherapy of L]210-bearing mice with L.J CD2F1 were challenged i.p. with different numbers of L1210 cells (day 0). Experimental animals were given injections of L. I (ascitic fluid containing 30mg MAbml-1) alone or combined with rabbit complement, starting 24h after challenge. Different schedules of treatment were adopted. In the combined therapy, L. 1 was given i.p. with cyclophosphamide 120 mg kg-1 s.c. Tumour growth was evaluated by recording the median survival time (MST) and the number of dead animals per group (D/T) of control and treated animals.
In vitro evaluation of serotherapj' In vivo binding of L.] to L1210 At different intervals after the i.v. infusion of L.1 into L1210bearing mice, tumour cells were collected from peritoneal cavity and divided into 2 pools. The first pool was analysed by IMF for the presence of L.1 on the cells surface. The cells from the second pool were labelled with 5Cr and assayed for their susceptibility to lysis by the CDC system.
Detection of L.J in the serum of normal mice Blood samples from CD2F, mice infused i.v. with L.1 were collected at different times after the injection and the serum tested for the presence of L.I by both the IMF and CDC systems.
Detection of circulating antigen in L1210 mouse fluids The presence of the antigen defined by L.1 in the serum and in the ascites of L1210-bearing mice was detected by incubating at 4°C for 45min an appropriate dilution of L. 1 (supernatant from hydridoma culture) with an equal volume of serum or ascites (undiluted, 1:2, 1:4 etc.) from mice at different days after tumour challenge. The resulting L. 1 cytotoxic activity was assayed against L 1210 target cells. Inhibition of complement activity by normal mouse serum Normal sera from different strains of mice were mixed for 45min at 4°C with 5Cr-labelled L1210 cells (106) and L.1 culture supernatant. The mixture (50,ul) was seeded in a 96-well microplate, complement was added (25 pl) and the cytotoxic activity of L. 1 was evaluated as reported above (CDC).

Effects of L.] administration on survival of leukaemic mice
The results of passive serotherapy of leukaemic mice with L. I and rabbit serum, as source of SEROTHERAPY OF MURINE LEUKAEMIA BY MONOCLIONAL ANTIBODIES whether the MAb acted synergistically with a chemotherapeutic agent, leukaemic mice were treated with an effective dose of cyclophosphamide plus L.1. As shown in Table III serotherapy did not significantly increase the survival time over cyclophosphamide alone.

L.1 binding to L1210 cells in vivo
In order to study the reasons for the failure of L. 1 therapy, experiments were designed to determine whether the L. 1 could bind specifically to tumour cells in vivo. Figure shows that, although 24 h after L. administration the majority of Ll 210 cells were stained, a remarkable proportion of unstained cells was still detectable. Moreover, the fluorescence intensity of stained cells was never >20% of that of L1210 cells treated in vitro with L.1. The % of + ve cells and fluorescence intensity were completely restored when Ll210 cells interacting with L. in vivo were further exposed to the MAb in vitro.
Results of time-course experiments are also illustrated in Figure 1. The % of L. 1 + ve cells analysed by automatic flow fluorocytometry peaked  24 h after the L. 1 inoculation and decreased gradually, although 15% of + ve cells were still detected 6 days after the treatment. As shown in Table IV, the amount of L.1 bound to L1210 cells after in vivo treatment was dependent upon the dose of the antibody injected. Findings in keeping with those reported in Figure 1 were obtained in the cytotoxic assay. (Figure 2). A % of tumour cells derived from L 1210-bearing mice previously treated with L. 1 underwent lysis upon in vitro incubationwith rabbit complement. However, full susceptibility to complement-mediated lysis was obtained upon cell re-exposure to L. I in vitro.

Studies of L. I properties after in vivo administration
The time course of blood L.1 levels in CD2F1 mice, as evaluated by IMF (dashed line) or by the 51Crrelease assay (full line), was studied ( Figure 3). look  L1210 cells incubated with serum from L. 1 treated mice were stained positively in the IMF assay. After 4 days, the binding activity dropped progressively, although it was still detectable 15 days after L. 1 inoculation. In sharp contrast, no cytotoxic activity was present in serum from L.1treated mice. The possibility that complement inactivation by mouse serum was responsible for the in vivo loss of L. 1 cytotoxic activity, was investigated. Serum samples from CD2F1 mice inoculated i.v. with rabbit complement did not show any complement activity in the CDC (data not shown). Moreover, the presence of undiluted mouse serum in the CDC assay inhibited the cytotoxic reaction, whatever the source of complement (Table V). However, a normal lytic reaction was observed when the mouse serum was removed before the addition of complement.

L1210 antigen in mousefluids
The detectability of antigen recognized by L.1 shed by tumour cells into the blood or the ascites of The cytotoxic activity of L.1 was evaluated in a CDC with L1210 target cells, in the presence of normal serum from different strains of mice as described in Materials and Methods. When indicated (w), L.1 and mouse serum were washed off before adding the complement. L.1 (supernatant from hybridoma culture) was present in all samples. Replacement of L.1 with rabbit anti mouse serum caused similar results.
*Sources of rabbit complement were selected from noncytotoxic blood samples from our breeding colony.
Guinea pig sources were purchased (Bio Mrieux, France). Human complement was from healthy donors. All samples were assayed in vitro for complement activity before use, and for each complement the appropriate dilution was selected. This Table uses rabbit complement as a + ve control.
L1210-bearing mice was examined. Inhibition of cytotoxic activity of L.1 by serum and ascites fluid from leukaemic mice obtained at varying intervals after the challenge is reported in Figure 4. The inhibitory capacity of mouse fluids, not detectable until 4 days after the tumour inoculation, increased rapidly thereafter and was more impressive for the ascites fluid than for the serum. Titration of soluble antigen ( Figure 5) showed that a relatively large amount of free antigen was detectable in the ascites fluid on day 7, whereas a lesser degree of inhibitory activity was demonstrable on day 5. Inhibition of the cytotoxic activity of L.1 by ascites and serum from L1210-bearing mice. The supernatant from the hybridoma culture was incubated at 4°C for 45 min with the serum (*-@) and the ascites (@----@) from L1210-bearing mice on different days after the challenge i.p. with 104 leukaemic cells. The mixture was added to L1210-labelled cells and incubated at 37°C (45 min). After washing an appropriate dilution of complement was added as described for the CDC assay. The results are expressed as % inhibition of L.1 cytotoxicy.

Discussion
In principle, specific and cytotoxic MAb should provide selective bullets for the in vivo eradication of cancerous cells leaving normal cells undamaged. These theoretical conditions, hypothesized since the Ehrlich era, were obtained in our laboratory by the production of L.1. In fact, L.1 is absolutely specific to L1210 leukaemic cells and as an IgM immunoglobulin displays great efficiency in fixing complement (Testorelli et al., 1982). Furthermore, since the CD2F1 mice used in these studies were of Inhibition of the cytotoxic activity of L.1 by cell-free ascites from L1210-bearing mice. Varying dilutions of cell-free ascites from L1210-bearing mice challenged i.p. with 104 leukaemic cells were incubated at 4°C for 45 min with the supernatant from the hybridoma culture. The residual cytotoxic activity of L.1 was determined in a CDC assay as reported in Figure 4. The inhibitory activity of the ascites obtained on Day 5 (0-0) and on Day 7 (x----x) after the L1210 challenge is illustrated.
identical haplotype to that of the L.1 hybridoma cells, strictly syngeneic conditions could be obtained. However, even under these optimal theoretical conditions, L.1 serotherapy was almost completely ineffective in increasing the survival time of L1210-bearing CD2F1 mice. Variations of the schedule of route or treatment, of the amounts of antibody or of complement were consistently unsuccessful. One reasonable cause for failure, the low level of endogenous complement in mice, was counteracted by excessive inoculation of exogenous complement, but this manoevre was without improvement. Since the activation of an endogenous immune response (Kirch & Hammerling, 1981) is a possible effector mechanism of serotherapy, L. 1 might synergize with an effective antitumour drug. Following this line, a number of treatments with cyclophosphamide or with other anticancer compounds (not reported here) were combined with L.1 serotherapy, without any additional increase in the life-span of the mouse.
To account for the failure of L. 1 serotherapy several experiments were devised. The first set was conceived to see whether or not L. 1 given i.v. to tumour-bearing mice bound to peritoneal L1210 cells. Although the results were positive (Figure 1) a lower percentage than expected of dead cells and of fluorescent cells was regularly observed. It is noteworthy that the fluorescence intensity of in vivo L. 1-treated cells was 80% below the intensity achieved in the in vitro-treated control cells. Decreased susceptibility to in vitro complementmediated lysis by L1210 cells exposed in vivo to L.1 was confirmed by the 5'Cr-release assay. Finally, we found that the susceptibility to both complement-mediated lysis and to staining with a second antibody were completely restored after in vitro re-exposure to L. 1.
The in vivo studies suggest the following conclusions: (a) a fraction of L1210 cells did not bind L.1 in vivo; (b) the amount of L.1 bound to the surface cells +ve by IMF-was less than for in vitro-treated controls, as demonstrated by the lower fluorescence intensity; (c) the in vivo L.1 +ve cells, in spite of their low fluorescence intensity, were still susceptible to in vitro lysis in presence of an appropriate source of complement; (d) the antigenic specificity defined by L. I did not undergo modulation, since fluorescence intensity and susceptibility to lysis were fully recovered after in vitro re-exposure to L.1: (e) cells -ve by IMF and those resistant to complement-mediated lysis might belong to the same subpopulation.
Although blood clearance of L.1 in the mouse had the half life already described by others for the same antibody class Kirch & Hammerling 1981), blood serum samples containing L.1 and complement did not lyse L1210 cells in vitro. Furthermore, complement from different sources mixed with mouse serum in vitro was completely inactivated.
Antigen shedding cannot be responsible for the failure of L. 1 serotherapy since, at the start of serotherapy, soluble antigen was not found in body fluids. L. 1 inhibition by mouse serum and even more by ascites fluid from tumour-bearing mice became evident 4 days after L1210 challenge. Therefore, circulating antigen might negatively influence late treatments only.
The experimental system used in these studies although adhering to optimal theoretical conditions, did not bring about increased survival time for leukaemic mice. Incomplete antibody binding to target cells and blood inactivation of complement might be major factors in these disappointing results.
A number of claims (Kirch & Hammerling, 1981;Bernstein et al., 1980;Scheinberg & Strand 1982;Miller & Levy, 1981;Ritz et al., 1980;Trowbridge & Lopez 1982;Herlyn & Koprowski, 1981), of therapeutic benefit from the use of MAb in cancer therapy have been reported recently in the literature. Partial successes in tumour systems other than that used here could be attributed to different Ig isotypes, acting through an antibody-dependent cellular cytotoxicity (ADCC) rather than complement-mediated lysis. The limited achievements of this approach might result in greater efforts (Pimm et al., 1982;Raso et al., 1982) to exploit monoclonal technology to carry toxic compounds specifically to their target cells.
Research supported in part by PFCCN, C.N.R., Rome, Italy.