Cox proportional hazards deep neural network identifies peripheral blood complete remission to be at least equivalent to morphologic complete remission in predicting outcomes of patients treated with azacitidine—A prospective cohort study by the AGMT
The current gold standard of response assessment in patients with myelodysplastic syndromes (MDS), chronic myelomonocytic leukemia (CMML), and acute myeloid leukemia (AML) is morphologic complete remission (CR) and CR with incomplete count recovery (CRi), both of which require an invasive BM evaluat...
Gespeichert in:
| Veröffentlicht in: | American journal of hematology 2023-11, Vol.98 (11), p.1685-1698 |
|---|---|
| Hauptverfasser: | , , , , , , , , , , , , , , , , , , , , , , , , |
| Format: | Artikel |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | 1698 |
|---|---|
| container_issue | 11 |
| container_start_page | 1685 |
| container_title | American journal of hematology |
| container_volume | 98 |
| creator | Pleyer, Lisa Vaisband, Marc Drost, Manuel Pfeilstöcker, Michael Stauder, Reinhard Heibl, Sonja Sill, Heinz Girschikofsky, Michael Stampfl‐Mattersberger, Margarete Pichler, Angelika Hartmann, Bernd Petzer, Andreas Schreder, Martin Schmitt, Clemens A. Vallet, Sonia Melchardt, Thomas Zebisch, Armin Pichler, Petra Zaborsky, Nadja Machherndl‐Spandl, Sigrid Wolf, Dominik Keil, Felix Hasenauer, Jan Larcher‐Senn, Julian Greil, Richard |
| description | The current gold standard of response assessment in patients with myelodysplastic syndromes (MDS), chronic myelomonocytic leukemia (CMML), and acute myeloid leukemia (AML) is morphologic complete remission (CR) and CR with incomplete count recovery (CRi), both of which require an invasive BM evaluation. Outside of clinical trials, BM evaluations are only performed in ~50% of patients during follow‐up, pinpointing a clinical need for response endpoints that do not necessitate BM assessments. We define and validate a new response type termed “peripheral blood complete remission” (PB‐CR) that can be determined from the differential blood count and clinical parameters without necessitating a BM assessment. We compared the predictive value of PB‐CR with morphologic CR/CRi in 1441 non‐selected, consecutive patients diagnosed with MDS (n = 522; 36.2%), CMML (n = 132; 9.2%), or AML (n = 787; 54.6%), included within the Austrian Myeloid Registry (aMYELOIDr; NCT04438889). Time‐to‐event analyses were adjusted for 17 covariates remaining in the final Cox proportional hazards (CPH) model. DeepSurv, a CPH neural network model, and permutation‐based feature importance were used to validate results. 1441 patients were included. Adjusted median overall survival for patients achieving PB‐CR was 22.8 months (95%CI 18.9–26.2) versus 10.4 months (95%CI 9.7–11.2) for those who did not; HR = 0.366 (95%CI 0.303–0.441; p |
| doi_str_mv | 10.1002/ajh.27046 |
| format | Article |
| fullrecord | <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_miscellaneous_2847341974</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2875462773</sourcerecordid><originalsourceid>FETCH-LOGICAL-c3886-d506a79a071aa27149eccd559afb78007bce1aa6c621d4710d604d1c8412c0743</originalsourceid><addsrcrecordid>eNp1ks9u1DAQxiMEokvhwAugkbjAYVs7_5wcVytoQUVcyjly7EnjJRu7ttNle-pD8IQ8Bwcm3cIBwcmW5zef59N8SfKSsxPOWHoqN_1JKlhePkoWnNXlsiqL9HGyYFnJ6c7qo-RZCBvGOM8r9jQ5ykSRV1nNFsnPtf0GzltnfTR2lAP08lZ6HUAjOhhx8vQ2YtxZ_xWMxjGazmAAh964HudqO1irQdmtGzAieNyaEEgMooUWQUYYUIYIeD2ZGzmQxFzZWu96O9gro_7Va0YaC7VR0YxXYKdIDH1rO3AyGtIIED3KiBp2JvZAUysTjTYj_rj7vpo9BYfUfYMk35M9CHHSe2j3EHuE1dmny-fJk04OAV88nMfJl_fvLtfny4vPZx_Wq4ulyqqqXOqClVLUkgkuZSp4XqNSuihq2bWiYky0CqlSqjLlOhec6ZLlmqsq56liIs-OkzcHXRrqesIQG3KpcBjkiHYKTVrlIst5fY--_gvd2MnTXmaKtlamQmREvT1QilwGj13jvNlKv284a-ZINBSJ5j4SxL56UJzaLeo_5O8MEHB6AHZmwP3_lZrVx_OD5C845shg</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2875462773</pqid></control><display><type>article</type><title>Cox proportional hazards deep neural network identifies peripheral blood complete remission to be at least equivalent to morphologic complete remission in predicting outcomes of patients treated with azacitidine—A prospective cohort study by the AGMT</title><source>Wiley Free Content</source><source>Wiley Online Library Journals Frontfile Complete</source><source>Elektronische Zeitschriftenbibliothek - Frei zugängliche E-Journals</source><creator>Pleyer, Lisa ; Vaisband, Marc ; Drost, Manuel ; Pfeilstöcker, Michael ; Stauder, Reinhard ; Heibl, Sonja ; Sill, Heinz ; Girschikofsky, Michael ; Stampfl‐Mattersberger, Margarete ; Pichler, Angelika ; Hartmann, Bernd ; Petzer, Andreas ; Schreder, Martin ; Schmitt, Clemens A. ; Vallet, Sonia ; Melchardt, Thomas ; Zebisch, Armin ; Pichler, Petra ; Zaborsky, Nadja ; Machherndl‐Spandl, Sigrid ; Wolf, Dominik ; Keil, Felix ; Hasenauer, Jan ; Larcher‐Senn, Julian ; Greil, Richard</creator><creatorcontrib>Pleyer, Lisa ; Vaisband, Marc ; Drost, Manuel ; Pfeilstöcker, Michael ; Stauder, Reinhard ; Heibl, Sonja ; Sill, Heinz ; Girschikofsky, Michael ; Stampfl‐Mattersberger, Margarete ; Pichler, Angelika ; Hartmann, Bernd ; Petzer, Andreas ; Schreder, Martin ; Schmitt, Clemens A. ; Vallet, Sonia ; Melchardt, Thomas ; Zebisch, Armin ; Pichler, Petra ; Zaborsky, Nadja ; Machherndl‐Spandl, Sigrid ; Wolf, Dominik ; Keil, Felix ; Hasenauer, Jan ; Larcher‐Senn, Julian ; Greil, Richard</creatorcontrib><description>The current gold standard of response assessment in patients with myelodysplastic syndromes (MDS), chronic myelomonocytic leukemia (CMML), and acute myeloid leukemia (AML) is morphologic complete remission (CR) and CR with incomplete count recovery (CRi), both of which require an invasive BM evaluation. Outside of clinical trials, BM evaluations are only performed in ~50% of patients during follow‐up, pinpointing a clinical need for response endpoints that do not necessitate BM assessments. We define and validate a new response type termed “peripheral blood complete remission” (PB‐CR) that can be determined from the differential blood count and clinical parameters without necessitating a BM assessment. We compared the predictive value of PB‐CR with morphologic CR/CRi in 1441 non‐selected, consecutive patients diagnosed with MDS (n = 522; 36.2%), CMML (n = 132; 9.2%), or AML (n = 787; 54.6%), included within the Austrian Myeloid Registry (aMYELOIDr; NCT04438889). Time‐to‐event analyses were adjusted for 17 covariates remaining in the final Cox proportional hazards (CPH) model. DeepSurv, a CPH neural network model, and permutation‐based feature importance were used to validate results. 1441 patients were included. Adjusted median overall survival for patients achieving PB‐CR was 22.8 months (95%CI 18.9–26.2) versus 10.4 months (95%CI 9.7–11.2) for those who did not; HR = 0.366 (95%CI 0.303–0.441; p < .0001). Among patients achieving CR, those additionally achieving PB‐CR had a median adjusted OS of 32.6 months (95%CI 26.2–49.2) versus 21.7 months (95%CI 16.9–27.7; HR = 0.400 [95%CI 0.190–0.844; p = .0161]) for those who did not. Our deep neural network analysis‐based findings from a large, prospective cohort study indicate that BM evaluations solely for the purpose of identifying CR/CRi can be omitted.
After adjusting for 17 baseline covariates, patients treated with azacitidine achieving peripheral blood complete remission (PB‐CR) had +12.4 months longer overall survival (OS) (22.8 vs. 10.4 months; p < .0001), and a 63% reduced risk of death (HR 0.366) than patients who did not achieve PB‐CR (Panel A). Among patients achieving morphologic CR, those additionally achieving PB‐CR had +10.9 months longer adjusted OS (32.6 vs. 21.7 months; p = .0161) and a 60% reduced risk of death (HR 0.400) (Panel B). After 5‐fold cross‐validation and Bonferroni correction for multiple testing, the validated deep neural network model DeepSurv (architecture depitcted in Panel C) and permutation‐based feature importances confirmed PB‐CR to have a higher information content than either CR or CR/CRi, and thus to be at least equivalent to these responses with regards to all time‐to‐event endpoints analyzed, including time on treatment, time to next treatment, and OS (p < .0001) (Panel D).</description><identifier>ISSN: 0361-8609</identifier><identifier>EISSN: 1096-8652</identifier><identifier>DOI: 10.1002/ajh.27046</identifier><identifier>PMID: 37548390</identifier><language>eng</language><publisher>Hoboken, USA: John Wiley & Sons, Inc</publisher><subject>Acute myeloid leukemia ; Chronic myelomonocytic leukemia ; Clinical trials ; Cohort analysis ; Hematology ; Leukemia ; Myelodysplastic syndrome ; Myelomonocytic leukemia ; Neural networks ; Peripheral blood ; Remission ; Remission (Medicine)</subject><ispartof>American journal of hematology, 2023-11, Vol.98 (11), p.1685-1698</ispartof><rights>2023 The Authors. published by Wiley Periodicals LLC.</rights><rights>2023 The Authors. American Journal of Hematology published by Wiley Periodicals LLC.</rights><rights>2023. This article is published under http://creativecommons.org/licenses/by-nc-nd/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c3886-d506a79a071aa27149eccd559afb78007bce1aa6c621d4710d604d1c8412c0743</citedby><cites>FETCH-LOGICAL-c3886-d506a79a071aa27149eccd559afb78007bce1aa6c621d4710d604d1c8412c0743</cites><orcidid>0000-0003-0993-4371 ; 0000-0002-2790-5067</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktopdf>$$Uhttps://onlinelibrary.wiley.com/doi/pdf/10.1002%2Fajh.27046$$EPDF$$P50$$Gwiley$$Hfree_for_read</linktopdf><linktohtml>$$Uhttps://onlinelibrary.wiley.com/doi/full/10.1002%2Fajh.27046$$EHTML$$P50$$Gwiley$$Hfree_for_read</linktohtml><link.rule.ids>314,776,780,1411,1427,27901,27902,45550,45551,46384,46808</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/37548390$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink></links><search><creatorcontrib>Pleyer, Lisa</creatorcontrib><creatorcontrib>Vaisband, Marc</creatorcontrib><creatorcontrib>Drost, Manuel</creatorcontrib><creatorcontrib>Pfeilstöcker, Michael</creatorcontrib><creatorcontrib>Stauder, Reinhard</creatorcontrib><creatorcontrib>Heibl, Sonja</creatorcontrib><creatorcontrib>Sill, Heinz</creatorcontrib><creatorcontrib>Girschikofsky, Michael</creatorcontrib><creatorcontrib>Stampfl‐Mattersberger, Margarete</creatorcontrib><creatorcontrib>Pichler, Angelika</creatorcontrib><creatorcontrib>Hartmann, Bernd</creatorcontrib><creatorcontrib>Petzer, Andreas</creatorcontrib><creatorcontrib>Schreder, Martin</creatorcontrib><creatorcontrib>Schmitt, Clemens A.</creatorcontrib><creatorcontrib>Vallet, Sonia</creatorcontrib><creatorcontrib>Melchardt, Thomas</creatorcontrib><creatorcontrib>Zebisch, Armin</creatorcontrib><creatorcontrib>Pichler, Petra</creatorcontrib><creatorcontrib>Zaborsky, Nadja</creatorcontrib><creatorcontrib>Machherndl‐Spandl, Sigrid</creatorcontrib><creatorcontrib>Wolf, Dominik</creatorcontrib><creatorcontrib>Keil, Felix</creatorcontrib><creatorcontrib>Hasenauer, Jan</creatorcontrib><creatorcontrib>Larcher‐Senn, Julian</creatorcontrib><creatorcontrib>Greil, Richard</creatorcontrib><title>Cox proportional hazards deep neural network identifies peripheral blood complete remission to be at least equivalent to morphologic complete remission in predicting outcomes of patients treated with azacitidine—A prospective cohort study by the AGMT</title><title>American journal of hematology</title><addtitle>Am J Hematol</addtitle><description>The current gold standard of response assessment in patients with myelodysplastic syndromes (MDS), chronic myelomonocytic leukemia (CMML), and acute myeloid leukemia (AML) is morphologic complete remission (CR) and CR with incomplete count recovery (CRi), both of which require an invasive BM evaluation. Outside of clinical trials, BM evaluations are only performed in ~50% of patients during follow‐up, pinpointing a clinical need for response endpoints that do not necessitate BM assessments. We define and validate a new response type termed “peripheral blood complete remission” (PB‐CR) that can be determined from the differential blood count and clinical parameters without necessitating a BM assessment. We compared the predictive value of PB‐CR with morphologic CR/CRi in 1441 non‐selected, consecutive patients diagnosed with MDS (n = 522; 36.2%), CMML (n = 132; 9.2%), or AML (n = 787; 54.6%), included within the Austrian Myeloid Registry (aMYELOIDr; NCT04438889). Time‐to‐event analyses were adjusted for 17 covariates remaining in the final Cox proportional hazards (CPH) model. DeepSurv, a CPH neural network model, and permutation‐based feature importance were used to validate results. 1441 patients were included. Adjusted median overall survival for patients achieving PB‐CR was 22.8 months (95%CI 18.9–26.2) versus 10.4 months (95%CI 9.7–11.2) for those who did not; HR = 0.366 (95%CI 0.303–0.441; p < .0001). Among patients achieving CR, those additionally achieving PB‐CR had a median adjusted OS of 32.6 months (95%CI 26.2–49.2) versus 21.7 months (95%CI 16.9–27.7; HR = 0.400 [95%CI 0.190–0.844; p = .0161]) for those who did not. Our deep neural network analysis‐based findings from a large, prospective cohort study indicate that BM evaluations solely for the purpose of identifying CR/CRi can be omitted.
After adjusting for 17 baseline covariates, patients treated with azacitidine achieving peripheral blood complete remission (PB‐CR) had +12.4 months longer overall survival (OS) (22.8 vs. 10.4 months; p < .0001), and a 63% reduced risk of death (HR 0.366) than patients who did not achieve PB‐CR (Panel A). Among patients achieving morphologic CR, those additionally achieving PB‐CR had +10.9 months longer adjusted OS (32.6 vs. 21.7 months; p = .0161) and a 60% reduced risk of death (HR 0.400) (Panel B). After 5‐fold cross‐validation and Bonferroni correction for multiple testing, the validated deep neural network model DeepSurv (architecture depitcted in Panel C) and permutation‐based feature importances confirmed PB‐CR to have a higher information content than either CR or CR/CRi, and thus to be at least equivalent to these responses with regards to all time‐to‐event endpoints analyzed, including time on treatment, time to next treatment, and OS (p < .0001) (Panel D).</description><subject>Acute myeloid leukemia</subject><subject>Chronic myelomonocytic leukemia</subject><subject>Clinical trials</subject><subject>Cohort analysis</subject><subject>Hematology</subject><subject>Leukemia</subject><subject>Myelodysplastic syndrome</subject><subject>Myelomonocytic leukemia</subject><subject>Neural networks</subject><subject>Peripheral blood</subject><subject>Remission</subject><subject>Remission (Medicine)</subject><issn>0361-8609</issn><issn>1096-8652</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>24P</sourceid><recordid>eNp1ks9u1DAQxiMEokvhwAugkbjAYVs7_5wcVytoQUVcyjly7EnjJRu7ttNle-pD8IQ8Bwcm3cIBwcmW5zef59N8SfKSsxPOWHoqN_1JKlhePkoWnNXlsiqL9HGyYFnJ6c7qo-RZCBvGOM8r9jQ5ykSRV1nNFsnPtf0GzltnfTR2lAP08lZ6HUAjOhhx8vQ2YtxZ_xWMxjGazmAAh964HudqO1irQdmtGzAieNyaEEgMooUWQUYYUIYIeD2ZGzmQxFzZWu96O9gro_7Va0YaC7VR0YxXYKdIDH1rO3AyGtIIED3KiBp2JvZAUysTjTYj_rj7vpo9BYfUfYMk35M9CHHSe2j3EHuE1dmny-fJk04OAV88nMfJl_fvLtfny4vPZx_Wq4ulyqqqXOqClVLUkgkuZSp4XqNSuihq2bWiYky0CqlSqjLlOhec6ZLlmqsq56liIs-OkzcHXRrqesIQG3KpcBjkiHYKTVrlIst5fY--_gvd2MnTXmaKtlamQmREvT1QilwGj13jvNlKv284a-ZINBSJ5j4SxL56UJzaLeo_5O8MEHB6AHZmwP3_lZrVx_OD5C845shg</recordid><startdate>202311</startdate><enddate>202311</enddate><creator>Pleyer, Lisa</creator><creator>Vaisband, Marc</creator><creator>Drost, Manuel</creator><creator>Pfeilstöcker, Michael</creator><creator>Stauder, Reinhard</creator><creator>Heibl, Sonja</creator><creator>Sill, Heinz</creator><creator>Girschikofsky, Michael</creator><creator>Stampfl‐Mattersberger, Margarete</creator><creator>Pichler, Angelika</creator><creator>Hartmann, Bernd</creator><creator>Petzer, Andreas</creator><creator>Schreder, Martin</creator><creator>Schmitt, Clemens A.</creator><creator>Vallet, Sonia</creator><creator>Melchardt, Thomas</creator><creator>Zebisch, Armin</creator><creator>Pichler, Petra</creator><creator>Zaborsky, Nadja</creator><creator>Machherndl‐Spandl, Sigrid</creator><creator>Wolf, Dominik</creator><creator>Keil, Felix</creator><creator>Hasenauer, Jan</creator><creator>Larcher‐Senn, Julian</creator><creator>Greil, Richard</creator><general>John Wiley & Sons, Inc</general><general>Wiley Subscription Services, Inc</general><scope>24P</scope><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>K9.</scope><scope>7X8</scope><orcidid>https://orcid.org/0000-0003-0993-4371</orcidid><orcidid>https://orcid.org/0000-0002-2790-5067</orcidid></search><sort><creationdate>202311</creationdate><title>Cox proportional hazards deep neural network identifies peripheral blood complete remission to be at least equivalent to morphologic complete remission in predicting outcomes of patients treated with azacitidine—A prospective cohort study by the AGMT</title><author>Pleyer, Lisa ; Vaisband, Marc ; Drost, Manuel ; Pfeilstöcker, Michael ; Stauder, Reinhard ; Heibl, Sonja ; Sill, Heinz ; Girschikofsky, Michael ; Stampfl‐Mattersberger, Margarete ; Pichler, Angelika ; Hartmann, Bernd ; Petzer, Andreas ; Schreder, Martin ; Schmitt, Clemens A. ; Vallet, Sonia ; Melchardt, Thomas ; Zebisch, Armin ; Pichler, Petra ; Zaborsky, Nadja ; Machherndl‐Spandl, Sigrid ; Wolf, Dominik ; Keil, Felix ; Hasenauer, Jan ; Larcher‐Senn, Julian ; Greil, Richard</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c3886-d506a79a071aa27149eccd559afb78007bce1aa6c621d4710d604d1c8412c0743</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Acute myeloid leukemia</topic><topic>Chronic myelomonocytic leukemia</topic><topic>Clinical trials</topic><topic>Cohort analysis</topic><topic>Hematology</topic><topic>Leukemia</topic><topic>Myelodysplastic syndrome</topic><topic>Myelomonocytic leukemia</topic><topic>Neural networks</topic><topic>Peripheral blood</topic><topic>Remission</topic><topic>Remission (Medicine)</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Pleyer, Lisa</creatorcontrib><creatorcontrib>Vaisband, Marc</creatorcontrib><creatorcontrib>Drost, Manuel</creatorcontrib><creatorcontrib>Pfeilstöcker, Michael</creatorcontrib><creatorcontrib>Stauder, Reinhard</creatorcontrib><creatorcontrib>Heibl, Sonja</creatorcontrib><creatorcontrib>Sill, Heinz</creatorcontrib><creatorcontrib>Girschikofsky, Michael</creatorcontrib><creatorcontrib>Stampfl‐Mattersberger, Margarete</creatorcontrib><creatorcontrib>Pichler, Angelika</creatorcontrib><creatorcontrib>Hartmann, Bernd</creatorcontrib><creatorcontrib>Petzer, Andreas</creatorcontrib><creatorcontrib>Schreder, Martin</creatorcontrib><creatorcontrib>Schmitt, Clemens A.</creatorcontrib><creatorcontrib>Vallet, Sonia</creatorcontrib><creatorcontrib>Melchardt, Thomas</creatorcontrib><creatorcontrib>Zebisch, Armin</creatorcontrib><creatorcontrib>Pichler, Petra</creatorcontrib><creatorcontrib>Zaborsky, Nadja</creatorcontrib><creatorcontrib>Machherndl‐Spandl, Sigrid</creatorcontrib><creatorcontrib>Wolf, Dominik</creatorcontrib><creatorcontrib>Keil, Felix</creatorcontrib><creatorcontrib>Hasenauer, Jan</creatorcontrib><creatorcontrib>Larcher‐Senn, Julian</creatorcontrib><creatorcontrib>Greil, Richard</creatorcontrib><collection>Wiley Online Library Open Access</collection><collection>PubMed</collection><collection>CrossRef</collection><collection>ProQuest Health & Medical Complete (Alumni)</collection><collection>MEDLINE - Academic</collection><jtitle>American journal of hematology</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Pleyer, Lisa</au><au>Vaisband, Marc</au><au>Drost, Manuel</au><au>Pfeilstöcker, Michael</au><au>Stauder, Reinhard</au><au>Heibl, Sonja</au><au>Sill, Heinz</au><au>Girschikofsky, Michael</au><au>Stampfl‐Mattersberger, Margarete</au><au>Pichler, Angelika</au><au>Hartmann, Bernd</au><au>Petzer, Andreas</au><au>Schreder, Martin</au><au>Schmitt, Clemens A.</au><au>Vallet, Sonia</au><au>Melchardt, Thomas</au><au>Zebisch, Armin</au><au>Pichler, Petra</au><au>Zaborsky, Nadja</au><au>Machherndl‐Spandl, Sigrid</au><au>Wolf, Dominik</au><au>Keil, Felix</au><au>Hasenauer, Jan</au><au>Larcher‐Senn, Julian</au><au>Greil, Richard</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Cox proportional hazards deep neural network identifies peripheral blood complete remission to be at least equivalent to morphologic complete remission in predicting outcomes of patients treated with azacitidine—A prospective cohort study by the AGMT</atitle><jtitle>American journal of hematology</jtitle><addtitle>Am J Hematol</addtitle><date>2023-11</date><risdate>2023</risdate><volume>98</volume><issue>11</issue><spage>1685</spage><epage>1698</epage><pages>1685-1698</pages><issn>0361-8609</issn><eissn>1096-8652</eissn><abstract>The current gold standard of response assessment in patients with myelodysplastic syndromes (MDS), chronic myelomonocytic leukemia (CMML), and acute myeloid leukemia (AML) is morphologic complete remission (CR) and CR with incomplete count recovery (CRi), both of which require an invasive BM evaluation. Outside of clinical trials, BM evaluations are only performed in ~50% of patients during follow‐up, pinpointing a clinical need for response endpoints that do not necessitate BM assessments. We define and validate a new response type termed “peripheral blood complete remission” (PB‐CR) that can be determined from the differential blood count and clinical parameters without necessitating a BM assessment. We compared the predictive value of PB‐CR with morphologic CR/CRi in 1441 non‐selected, consecutive patients diagnosed with MDS (n = 522; 36.2%), CMML (n = 132; 9.2%), or AML (n = 787; 54.6%), included within the Austrian Myeloid Registry (aMYELOIDr; NCT04438889). Time‐to‐event analyses were adjusted for 17 covariates remaining in the final Cox proportional hazards (CPH) model. DeepSurv, a CPH neural network model, and permutation‐based feature importance were used to validate results. 1441 patients were included. Adjusted median overall survival for patients achieving PB‐CR was 22.8 months (95%CI 18.9–26.2) versus 10.4 months (95%CI 9.7–11.2) for those who did not; HR = 0.366 (95%CI 0.303–0.441; p < .0001). Among patients achieving CR, those additionally achieving PB‐CR had a median adjusted OS of 32.6 months (95%CI 26.2–49.2) versus 21.7 months (95%CI 16.9–27.7; HR = 0.400 [95%CI 0.190–0.844; p = .0161]) for those who did not. Our deep neural network analysis‐based findings from a large, prospective cohort study indicate that BM evaluations solely for the purpose of identifying CR/CRi can be omitted.
After adjusting for 17 baseline covariates, patients treated with azacitidine achieving peripheral blood complete remission (PB‐CR) had +12.4 months longer overall survival (OS) (22.8 vs. 10.4 months; p < .0001), and a 63% reduced risk of death (HR 0.366) than patients who did not achieve PB‐CR (Panel A). Among patients achieving morphologic CR, those additionally achieving PB‐CR had +10.9 months longer adjusted OS (32.6 vs. 21.7 months; p = .0161) and a 60% reduced risk of death (HR 0.400) (Panel B). After 5‐fold cross‐validation and Bonferroni correction for multiple testing, the validated deep neural network model DeepSurv (architecture depitcted in Panel C) and permutation‐based feature importances confirmed PB‐CR to have a higher information content than either CR or CR/CRi, and thus to be at least equivalent to these responses with regards to all time‐to‐event endpoints analyzed, including time on treatment, time to next treatment, and OS (p < .0001) (Panel D).</abstract><cop>Hoboken, USA</cop><pub>John Wiley & Sons, Inc</pub><pmid>37548390</pmid><doi>10.1002/ajh.27046</doi><tpages>14</tpages><orcidid>https://orcid.org/0000-0003-0993-4371</orcidid><orcidid>https://orcid.org/0000-0002-2790-5067</orcidid><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext |
| identifier | ISSN: 0361-8609 |
| ispartof | American journal of hematology, 2023-11, Vol.98 (11), p.1685-1698 |
| issn | 0361-8609 1096-8652 |
| language | eng |
| recordid | cdi_proquest_miscellaneous_2847341974 |
| source | Wiley Free Content; Wiley Online Library Journals Frontfile Complete; Elektronische Zeitschriftenbibliothek - Frei zugängliche E-Journals |
| subjects | Acute myeloid leukemia Chronic myelomonocytic leukemia Clinical trials Cohort analysis Hematology Leukemia Myelodysplastic syndrome Myelomonocytic leukemia Neural networks Peripheral blood Remission Remission (Medicine) |
| title | Cox proportional hazards deep neural network identifies peripheral blood complete remission to be at least equivalent to morphologic complete remission in predicting outcomes of patients treated with azacitidine—A prospective cohort study by the AGMT |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-29T21%3A06%3A08IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Cox%20proportional%20hazards%20deep%20neural%20network%20identifies%20peripheral%20blood%20complete%20remission%20to%20be%20at%20least%20equivalent%20to%20morphologic%20complete%20remission%20in%20predicting%20outcomes%20of%20patients%20treated%20with%20azacitidine%E2%80%94A%20prospective%20cohort%20study%20by%20the%20AGMT&rft.jtitle=American%20journal%20of%20hematology&rft.au=Pleyer,%20Lisa&rft.date=2023-11&rft.volume=98&rft.issue=11&rft.spage=1685&rft.epage=1698&rft.pages=1685-1698&rft.issn=0361-8609&rft.eissn=1096-8652&rft_id=info:doi/10.1002/ajh.27046&rft_dat=%3Cproquest_cross%3E2875462773%3C/proquest_cross%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2875462773&rft_id=info:pmid/37548390&rfr_iscdi=true |