Enhanced Antitumor Immunity Elicited by Dendritic Cell Vaccines Is a Result of Their Ability to Engage Both CTL and IFNγ-producing NK Cells
Bibliographic record
Abstract
Activation of cytotoxic T lymphocytes (CTLs) is a primary goal of many cancer vaccination therapies. We have evaluated two vaccination platforms, dendritic cells (DCs) and recombinant adenoviruses (rAds), for their ability to elicit CTL response and antitumor protection. Although rAd was more potent in CTL priming, DC vaccination provided greater protective and therapeutic antitumor activity. Subsequent analyses ruled out the possibility that the two vaccines elicit qualitatively distinct CTL, and demonstrated instead that DCs could better engage natural killer (NK) cells as an additional effector mechanism. We demonstrated that, although both DCs and rAd can stimulate rapid NK expansion, only DC-activated NK cells are able to produce interferon-γ (IFNγ) and mediate antitumor protection. Moreover, our data showed that exogenously delivered DCs preferentially engaged the Mac-1highCD27high NK subset, thereby suggesting that this NK population plays a predominant role in NK:DC interaction. Interestingly, at least 3 days were required for DC-triggered NK cells to acquire effector functions, indicating that a similar priming process operates between T cells and NK cells. Our results suggest that the nature of the vaccine platform can determine the relative involvement of NK and T cells in antitumor immunity, and that methods to augment NK function should be included in vaccination strategies in order to complement CTL-mediated control of tumor growth. Activation of cytotoxic T lymphocytes (CTLs) is a primary goal of many cancer vaccination therapies. We have evaluated two vaccination platforms, dendritic cells (DCs) and recombinant adenoviruses (rAds), for their ability to elicit CTL response and antitumor protection. Although rAd was more potent in CTL priming, DC vaccination provided greater protective and therapeutic antitumor activity. Subsequent analyses ruled out the possibility that the two vaccines elicit qualitatively distinct CTL, and demonstrated instead that DCs could better engage natural killer (NK) cells as an additional effector mechanism. We demonstrated that, although both DCs and rAd can stimulate rapid NK expansion, only DC-activated NK cells are able to produce interferon-γ (IFNγ) and mediate antitumor protection. Moreover, our data showed that exogenously delivered DCs preferentially engaged the Mac-1highCD27high NK subset, thereby suggesting that this NK population plays a predominant role in NK:DC interaction. Interestingly, at least 3 days were required for DC-triggered NK cells to acquire effector functions, indicating that a similar priming process operates between T cells and NK cells. Our results suggest that the nature of the vaccine platform can determine the relative involvement of NK and T cells in antitumor immunity, and that methods to augment NK function should be included in vaccination strategies in order to complement CTL-mediated control of tumor growth. IntroductionMost cancer vaccines have aimed at eliciting CD8+ cytotoxic T lymphocyte (CTL) activity because CTLs have the ability to directly eliminate tumor cells that express specific antigens. In this regard, both the quantity and quality (i.e., functional avidity) of antigen-specific CTLs are critical to the efficacy of antitumor responses.1Yang S Tsang KY Schlom J Induction of higher-avidity human CTLs by vector-mediated enhanced costimulation of antigen-presenting cells.Clin Cancer Res. 2005; 11: 5603-5615Crossref PubMed Scopus (33) Google Scholar,2Zeh HJ 3rd Perry-Lalley D Dudley ME Rosenberg SA Yang JC High avidity CTLs for two self-antigens demonstrate superior in vitro and in vivo antitumor efficacy.J Immunol. 1999; 162: 989-994PubMed Google Scholar However, tumors often express low levels of major histocompatibility complex class I molecules, thereby enabling their escape from CTL-mediated elimination. Ultimate control of tumor growth will therefore require additional effector mechanisms.3Gilboa E How tumors escape immune destruction and what we can do about it.Cancer Immunol Immunother. 1999; 48: 382-385Crossref PubMed Scopus (123) Google Scholar Studies have shown that natural killer (NK) cells can spontaneously detect and lyse transformed cells, possibly as a result of low major histocompatibility complex class I expression, thereby controlling tumor initiation and spread.4Mocikat R Braumuller J Gumy A Egeter O Ziegler H Reusch U et al.Natural killer cells activated by MHC class I(low) targets prime dendritic cells to induce protective CD8 T cell responses.Immunity. 2003; 19: 561-569Abstract Full Text Full Text PDF PubMed Scopus (342) Google Scholar,5Glas R Franksson L Une C Eloranta M Ohlen C Orn A et al.Recruitment and activation of natural killer (NK) cells in vivo determined by the target cell phenotype. An adaptive component of NK cell-mediated responses.J Exp Med. 2000; 191: 129-138Crossref PubMed Scopus (132) Google Scholar Additionally, NK cells are able to kill tumors that express normal levels of major histocompatibility complex class I, by activating the receptors that recognize specific ligands expressed by many tumor cells.6Cerwenka A Baron JL Lanier LL Ectopic expression of retinoic acid early inducible-1 gene (RAE-1) permits natural killer cell-mediated rejection of a MHC class I-bearing tumor in vivo.Proc Natl Acad Sci USA. 2001; 98: 11521-11526Crossref PubMed Scopus (493) Google Scholar Furthermore, upon activation, NK cells secrete cytokines such as interferon-γ (IFNγ) that can modulate both innate and adaptive antitumor activities.7Mailliard RB Son YI Redlinger R Coates PT Giermasz A Morel PA et al.Dendritic cells mediate NK cell help for Th1 and CTL responses: two-signal requirement for the induction of NK cell helper function.J Immunol. 2003; 171: 2366-2373Crossref PubMed Scopus (309) Google Scholar,8Martin-Fontecha A Thomsen LL Brett S Gerard C Lipp M Lanzavecchia A et al.Induced recruitment of NK cells to lymph nodes provides IFN-gamma for T(H)1 priming.Nat Immunol. 2004; 5: 1260-1265Crossref PubMed Scopus (1100) Google Scholar Taken together, these data suggest important roles for both CTL and NK cells in antitumor immunity.Two vaccination platforms, dendritic cells (DCs) and recombinant adenoviruses (rAds), have been intensively evaluated in animal models as as in A C therapeutic vaccination Immunol. 2005; PubMed Scopus Google S of dendritic cells as a to induce Immunol. Google Scholar DCs are important for the induction of a immune because have a to cell and CD8+ T cells. can mediate expression in a of cell in thereby to antigen-specific CD8+ S of dendritic cells as a to induce Immunol. Google J of the CD8+ response 2003; PubMed Scopus Google Scholar Moreover, have shown that both DCs and can engage NK cells of the tumor the for cancer vaccines to target effector H Yang H et and cells to tumor in the Immunol. PubMed Scopus Google A stimulate natural killer cell and demonstrate antitumor in the of 5: Full Text Full Text PDF PubMed Scopus Google E activation of adaptive and innate antitumor Immunol. 2003; 171: PubMed Scopus Google Scholar We have in tumor that vaccines are more potent in eliciting CTL vaccines DC vaccination provides better antitumor J R J cells an a for tumor PubMed Scopus Google J A J dendritic cells prime T cells a of and for cancer vaccines Cancer Res. 2000; Google J C et and antitumor is determined by the tumor the Immunol. 2004; PubMed Scopus Google Scholar was the is by the quality of the CTLs by the two vaccination platforms, as by et E functional and of CTL recombinant Immunol. 2005; PubMed Scopus Google Scholar by the involvement of NK cells as an additional effector a of primary CTL by DC and rAd and a similar quality of CTL However, the two in their to stimulate NK functional DCs and could induce rapid and activation of NK cells in the lymph the of and activated NK cells was and more in as a result of In the of specific tumor antitumor was only in in an analyses that NK cells more and the NK cells thereby suggesting that NK cells by DC have an additional process of functional We that the superior antitumor by DC vaccination was of the of antigen-specific CTL and activated NK and vaccines elicit qualitatively CD8+ the of and in a in were of cells days vaccines provided a of the that were a control DCs tumors in thereby that antitumor was specific However, were of tumor cells by was the of rAd to protective efficacy and data In of the cells and more the superior efficacy of DC vaccination was in a therapeutic the rAd the DC vaccine was days of cells that the levels of antitumor by and DCs be by in the levels of CTL both and that the of the antigen-specific CD8+ response by was greater that by In vivo CTL demonstrated that of target cells was greater in the was to to the quantity of CTL However, that the CD8+ T cells by both vaccines to the of in of thereby suggesting functional Moreover, the of CTL response was in and this ruled out the possibility that CD8+ T cell activated by DCs more of of antigen-specific CD8+ T cells from provided control of tumor growth Taken together, these data suggest that, in to CTL, to the protective response by DC T lymphocyte (CTL) by dendritic cell and rAd vaccines do in of CD8+ T cells was determined by and interferon-γ (IFNγ) data from are from target cells were for the in vivo CTL were target cell and the of antigen-specific was determined by from at this from were for of and for CD8 and to determine functional avidity of antigen-specific CD8+ T cells. data from days were the of CD8+ T cells in the was determined by of in of CD8+ T cells from the and lymph nodes of were that been cells were tumor and tumor the of were greater and activation of NK cells in the lymph rAd and DC vaccines are of activating NK cells, could the of antitumor H Yang H et and cells to tumor in the Immunol. PubMed Scopus Google A stimulate natural killer cell and demonstrate antitumor in the of 5: Full Text Full Text PDF PubMed Scopus Google E activation of adaptive and innate antitumor Immunol. 2003; 171: PubMed Scopus Google Scholar In order to determine the relative of the two vaccines in NK cells, we the of NK and activation in the of specific by control rAd Although a similar of in the lymph nodes was rAd DC both the of NK cells and the of NK cells were greater in showed that, DC NK was more days of the response is more potent dendritic cells (DCs) in natural killer (NK) and were DC lymph nodes were the days and for NK cell of cells from two lymph nodes was from DC and at of NK cells in the lymph nodes was by cell Activation of NK cells was by for of NK cells was determined by for and DC data from 3 to cells activated by DCs and order to determine the of NK cells, we by NK cells from at was in by NK cells at from to days In the of NK cells to in and for at least days suggest that DC provides additional to the functional of NK killer (NK) cells activated by dendritic cells (DCs) interferon-γ (IFNγ) and were DC the lymphocytes were at the and for a in and the was similar were DC recombinant and was out as NK for and days DC rAd were cells. cells were from and for and NK results shown are of NK cells were cells in and the of was a for and in NK cells was in the of that DC in the days and thereby that the of NK activation was a function to the two and of NK cells were from the of tumor thereby suggesting recruitment of NK cells to the tumor In vitro showed that NK cells of tumor cells, NK cells greater these results that DCs have a to the functional of NK cells, both for and for elicit and by NK distinct of NK cells have been in both and of human natural Immunol. 2001; Full Text Full Text PDF PubMed Scopus Google NK cells two distinct and Immunol. PubMed Scopus Google Scholar Mac-1highCD27high NK cells produce greater of upon in vitro as to their NK cells two distinct and Immunol. PubMed Scopus Google Scholar In order to determine vaccine preferentially specific NK we the of NK cells in days DC rAd was shown to the of the NK population by and by a of the was by DC by and by that of the NK cells and express of specific thereby suggesting the possibility that these a role in rapid of However, only NK cells were of thereby indicating that is Although could be out by of the NK cell and of the NK cells to by the two thereby the that NK cells a predominant role in NK:DC NK cells two distinct and Immunol. PubMed Scopus Google cell preferentially and interferon-γ (IFNγ) of Mac-1highCD27high natural killer (NK) were days DC recombinant and for the cell and for and data shown are of NK cells potent to determine NK cells activated by DC rAd in their ability to protective tumor were a of days to cells. was to in the of rAd and were was by in vivo of thereby suggesting that were effector We this in and by DC thereby out the possibility that cells are for protection. Moreover, of cells in these the that NK cells were the primary effector population cell vaccination provides natural killer antitumor and were DCs for days and cells. NK tumor and for two were days and tumor were at the a results are of from were DCs and cells at the were days for the of tumor were and cells days In vivo of CD8 NK cells tumor and for 3 of tumor was for order to out the possibility that rAd and DC NK cells were DC and cells at tumor was tumor cells were days DC In of the between rAd and tumor Interestingly, antitumor was in DC 3 days thereby suggesting that a process is for a functional of NK cells engaged by CTL and NK cells are required for protective in order to the better provided by the vaccine was to the of both CTL and NK cells, we in vivo specific and to a of cells. the results shown in of were CD8+ T cells and NK cells were required for this protective of of these to tumor the protection. T cells an effector role in this because are T cell that are to both the tumor and our data demonstrate the between CD8+ T cells and NK cells in the antitumor effector engaged by DC have evaluated two vaccination platforms, DC and for their ability to elicit antitumor in as as therapeutic Our analyses of CTL and NK cell a for an between antitumor efficacy and effector cell by DC and rAd have demonstrated that can the and of CD8+ T E functional and of CTL recombinant Immunol. 2005; PubMed Scopus Google A CD8+ and response Med. 2005; 11: PubMed Scopus Google J T R S et CD8+ T cell population by recombinant a that provides Immunol. PubMed Scopus Google Scholar and this our to determine the between the quality of CTL and antitumor by DC and rAd Our results that rAd and DC vaccines are in the rejection of cells at a and analyses demonstrate that CTLs by the two vaccine have functional the that rAd is DC tumor cell be by a of of CTL the between the CTL response and antitumor efficacy that been in et between and immune response Immunol. Google E J S M et T cell vaccination to in vitro to tumor Immunol. 1999; Google Scholar is that CTL be to control rapid tumor antitumor immune by cells and Cancer Res. PubMed Scopus Google H of in Med. 2004; PubMed Scopus Google Scholar of effector NK cells, will be for T antitumor a et have demonstrated that activating both CTL and NK cells of the activating can augment antitumor in as as therapeutic H M et vaccines innate and adaptive antitumor by the Natl Acad Sci USA. 2005; PubMed Scopus Google cells a population of cells a of activating and the these NK cells and functional LL NK cell Immunol. 2005; PubMed Scopus Google Scholar been that both and DCs can NK cells of tumor and to antitumor A stimulate natural killer cell and demonstrate antitumor in the of 5: Full Text Full Text PDF PubMed Scopus Google E activation of adaptive and innate antitumor Immunol. 2003; 171: PubMed Scopus Google Scholar However, their relative in NK functional and the by NK cells, are rAd is because of in the and is and of innate immune J Yang immune response to is by both and PubMed Scopus Google Scholar is therefore that greater and expression of NK cells are in the lymph nodes rAd DC However, a is that DC-activated NK cells a of and tumor cells. Furthermore, only were a in the of specific thereby suggesting that NK cells engaged by DC and rAd are our et have demonstrated that rAd in the of therapeutic expression to prime NK cells for antitumor A stimulate natural killer cell and demonstrate antitumor in the of 5: Full Text Full Text PDF PubMed Scopus Google Scholar However, of rAd at the of can tumor growth in an thereby suggesting that between NK and target cells for A stimulate natural killer cell and demonstrate antitumor in the of 5: Full Text Full Text PDF PubMed Scopus Google Scholar In of this a showed that of rAd human is the induction of NK cell-mediated in rAd results in of ligands target cells, NK activation and of the T et natural killer cell PubMed Scopus Google Scholar In our that DCs are able to NK cells in lymph in the of effector NK cells in the tumor these NK cells can be to mediate tumor Although the NK activation by the DC vaccine to be both by DC and NK have been in as T M L E cells and dendritic 2005; PubMed Scopus Google Scholar In et have provided that NK cell priming upon the of I by DCs and the and of by DCs to NK M A cells prime natural killer cells by Full Text Full Text PDF PubMed Scopus Google Scholar our these demonstrate that NK cells are activation and functional of NK cells require the of NK cells and a DC in and are that NK cells are more to lymphocytes of the adaptive immune been and require only a of for functional NK functional activation to by cells T cells that modulate NK cell J T cells NK cell-mediated of Immunol. PubMed Scopus Google C M C J et T cells natural killer cell in a growth Exp Med. 2005; PubMed Scopus Google Scholar Studies are in our to determine these can is that of DC rAd results in the of NK in vitro have that, human NK cells activated in response to DC only a is of and ME Activation and costimulation of human natural killer PubMed Scopus Google Scholar to the NK cells, for of the NK L C M D et and natural cell PubMed Scopus Google Scholar the of the of expression in et have provided that, as an NK cell expression can the NK cell two distinct NK cells two distinct and Immunol. PubMed Scopus Google Scholar are Mac-1highCD27high NK cells a greater ability to produce NK cells thereby suggesting that these NK cells a role in in NK cells two distinct and Immunol. PubMed Scopus Google Scholar Our results the in vivo that DC Mac-1highCD27high NK cell expansion, and thereby although and are to secrete In rAd induce NK for their we have intensively the of DC and rAd vaccines in eliciting antigen-specific CD8+ and innate NK activity. Although rAd vaccination is in CD8+ T cell is DC vaccination in NK functional that activation of both CTL and NK cells by DCs is for better antitumor possibly major histocompatibility class I low tumor cells by CTL-mediated to about control of rapid tumor R Braumuller J Gumy A Egeter O Ziegler H Reusch U et al.Natural killer cells activated by MHC class I(low) targets prime dendritic cells to induce protective CD8 T cell responses.Immunity. 2003; 19: 561-569Abstract Full Text Full Text PDF PubMed Scopus (342) Google Scholar,5Glas R Franksson L Une C Eloranta M Ohlen C Orn A et al.Recruitment and activation of natural killer (NK) cells in vivo determined by the target cell phenotype. An adaptive component of NK cell-mediated responses.J Exp Med. 2000; 191: 129-138Crossref PubMed Scopus (132) Google Scholar results have important only for vaccination strategies at NK cell effector functions, for a to the NK activation in response to and cell and to were from the were provided by of out were by the and the by the cells are from a cell M the in vivo protective Med. PubMed Scopus Google Scholar cells were in and is a that the of that been J of the CD8+ response 2003; PubMed Scopus Google Scholar is a that rAd were in cells and a as M J efficacy of vaccines in 2004; PubMed Scopus Google and were from and was from were from the at the were for to and and data were a and from cells were in the of for days as S et for of dendritic cells from Immunol 1999; PubMed Scopus Google Scholar cells were and days 3 and of the DCs were by of In the DCs were of for of and to and tumor DC and rAd vaccines were by were the of our their ability to induce CTL was J R J cells an a for tumor PubMed Scopus Google J of dendritic cytokines in immune 2001; PubMed Scopus Google Scholar were by tumor of cells therapeutic were an cells days to DC rAd In vivo of NK T cells was were tumor and control were for of of specific lymphocytes in was as by of tumors were for and the were a days tumor were days and the of antigen-specific CD8+ T cells was determined by and as J of the CD8+ response 2003; PubMed Scopus Google vivo CTL from were two was for at and of was was of cells from were and target cell were for of activity by cell was determined as a by the by and the by et R of CTL from lymph to Immunol. PubMed Scopus Google in CD8+ T cells. were for as L JL S A et al.Dendritic is required for the of cytotoxic CD8+ CTL in Immunol. 2001; PubMed Scopus (33) Google Scholar were and the from in the cells were in functional avidity of CD8+ T cells was determined by for of to We the of CD8+ T cells at as the response our of and in NK cells. cells were from of and for additional and the cells were for and of and to lymphocytes were for and for the in the of CD8+ T cells. and lymphocytes were from for CD8+ T cells were by and in order to determine the of antigen-specific effector cells. of CD8+ T cells from were in CD8+ T cells from a and that been cells were days NK activity was as A M M et of as the of tumor cells in a Immunol. 2001; PubMed Scopus Google Scholar was because the the that was required to of cells. NK cells were by and cells at a of for tumor cells were to and were and of tumor cells acid in was at were and of and was for indicating analyses were out IntroductionMost cancer vaccines have aimed at eliciting CD8+ cytotoxic T lymphocyte (CTL) activity because CTLs have the ability to directly eliminate tumor cells that express specific antigens. In this regard, both the quantity and quality (i.e., functional avidity) of antigen-specific CTLs are critical to the efficacy of antitumor responses.1Yang S Tsang KY Schlom J Induction of higher-avidity human CTLs by vector-mediated enhanced costimulation of antigen-presenting cells.Clin Cancer Res. 2005; 11: 5603-5615Crossref PubMed Scopus (33) Google Scholar,2Zeh HJ 3rd Perry-Lalley D Dudley ME Rosenberg SA Yang JC High avidity CTLs for two self-antigens demonstrate superior in vitro and in vivo antitumor efficacy.J Immunol. 1999; 162: 989-994PubMed Google Scholar However, tumors often express low levels of major histocompatibility complex class I molecules, thereby enabling their escape from CTL-mediated elimination. Ultimate control of tumor growth will therefore require additional effector mechanisms.3Gilboa E How tumors escape immune destruction and what we can do about it.Cancer Immunol Immunother. 1999; 48: 382-385Crossref PubMed Scopus (123) Google Scholar Studies have shown that natural killer (NK) cells can spontaneously detect and lyse transformed cells, possibly as a result of low major histocompatibility complex class I expression, thereby controlling tumor initiation and spread.4Mocikat R Braumuller J Gumy A Egeter O Ziegler H Reusch U et al.Natural killer cells activated by MHC class I(low) targets prime dendritic cells to induce protective CD8 T cell responses.Immunity. 2003; 19: 561-569Abstract Full Text Full Text PDF PubMed Scopus (342) Google Scholar,5Glas R Franksson L Une C Eloranta M Ohlen C Orn A et al.Recruitment and activation of natural killer (NK) cells in vivo determined by the target cell phenotype. An adaptive component of NK cell-mediated responses.J Exp Med. 2000; 191: 129-138Crossref PubMed Scopus (132) Google Scholar Additionally, NK cells are able to kill tumors that express normal levels of major histocompatibility complex class I, by activating the receptors that recognize specific ligands expressed by many tumor cells.6Cerwenka A Baron JL Lanier LL Ectopic expression of retinoic acid early inducible-1 gene (RAE-1) permits natural killer cell-mediated rejection of a MHC class I-bearing tumor in vivo.Proc Natl Acad Sci USA. 2001; 98: 11521-11526Crossref PubMed Scopus (493) Google Scholar Furthermore, upon activation, NK cells secrete cytokines such as interferon-γ (IFNγ) that can modulate both innate and adaptive antitumor activities.7Mailliard RB Son YI Redlinger R Coates PT Giermasz A Morel PA et al.Dendritic cells mediate NK cell help for Th1 and CTL responses: two-signal requirement for the induction of NK cell helper function.J Immunol. 2003; 171: 2366-2373Crossref PubMed Scopus (309) Google Scholar,8Martin-Fontecha A Thomsen LL Brett S Gerard C Lipp M Lanzavecchia A et al.Induced recruitment of NK cells to lymph nodes provides IFN-gamma for T(H)1 priming.Nat Immunol. 2004; 5: 1260-1265Crossref PubMed Scopus (1100) Google Scholar Taken together, these data suggest important roles for both CTL and NK cells in antitumor immunity.Two vaccination platforms, dendritic cells (DCs) and recombinant adenoviruses (rAds), have been intensively evaluated in animal models as as in A C therapeutic vaccination Immunol. 2005; PubMed Scopus Google S of dendritic cells as a to induce Immunol. Google Scholar DCs are important for the induction of a immune because have a to cell and CD8+ T cells. can mediate expression in a of cell in thereby to antigen-specific CD8+ S of dendritic cells as a to induce Immunol. Google J of the CD8+ response 2003; PubMed Scopus Google Scholar Moreover, have shown that both DCs and can engage NK cells of the tumor the for cancer vaccines to target effector H Yang H et and cells to tumor in the Immunol. PubMed Scopus Google A stimulate natural killer cell and demonstrate antitumor in the of 5: Full Text Full Text PDF PubMed Scopus Google E activation of adaptive and innate antitumor Immunol. 2003; 171: PubMed Scopus Google Scholar We have in tumor that vaccines are more potent in eliciting CTL vaccines DC vaccination provides better antitumor J R J cells an a for tumor PubMed Scopus Google J A J dendritic cells prime T cells a of and for cancer vaccines Cancer Res. 2000; Google J C et and antitumor is determined by the tumor the Immunol. 2004; PubMed Scopus Google Scholar was the is by the quality of the CTLs by the two vaccination platforms, as by et E functional and of CTL recombinant Immunol. 2005; PubMed Scopus Google Scholar by the involvement of NK cells as an additional effector a of primary CTL by DC and rAd and a similar quality of CTL However, the two in their to stimulate NK functional DCs and could induce rapid and activation of NK cells in the lymph the of and activated NK cells was and more in as a result of In the of specific tumor antitumor was only in in an analyses that NK cells more and the NK cells thereby suggesting that NK cells by DC have an additional process of functional We that the superior antitumor by DC vaccination was of the of antigen-specific CTL and activated NK cells.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".