{"paper_id":"1633ea8c-538a-4244-8ffa-6e55fcdeffa1","body_text":"Different flavonoid derivatives ( 1 – 37 ) were evaluated for cytotoxicity in endometrial (Ishikawa)\nand endometriotic (12Z) cell models.\nFlavonoids with hydroxyl/methoxy substituents, high\nTPSA, and multiple H-bond donors/acceptors showed low cytotoxicity\nand cytoprotective effects.\nFlavonoids\ncarrying electron-withdrawing substituents\n(−NO 2  and −Cl) and increased lipophilicity\n(log  P  > 3.5) displayed selective toxicity toward\n12Z cells.\nCompounds  24  [( E )-3-(4-(dimethylamino)­phenyl)­1-(3-hydroxyphenyl)­prop-2-en-1-one]\nand  28  [( E )-3-(benzo­[ d ]­[1,3]­dioxol-5-yl)-1-(4-chlorophenyl)­prop-2-en-1-one] emerged as\nthe most selective agents, reducing 12Z viability to ∼50% while\npreserving or enhancing Ishikawa viability.\nStructure–activity relationships suggested that\nlipophilicity contributes to selective toxicity, positioning chalcones\nas promising candidates for endometriosis therapy.\n\nEndometriosis is a benign\ngynecological disorder with a multifactorial\netiology involving genetic, epigenetic, immunological, hormonal, and\nenvironmental factors. These mechanisms collectively contribute to\nchronic inflammation and oxidative stress arising from the presence\nof ectopic endometrial tissues. The disease can be classified into\nthree major forms: endometrioma, deep endometriosis, and superficial\nperitoneal endometriosis. \n , \n  The incidence of endometriosis\naffects up to 20% of women of reproductive age, with common clinical\nmanifestations including dysmenorrhea, chronic pelvic pain, abnormal\nbleeding, and infertility. \n −\nCurrent therapeutic approaches typically combine\nlaparoscopic surgery\nwith hormonal treatments that aim at lesion removal and pain relief.\nHowever, these strategies often fail to restore fertility or prevent\ndisease recurrence. Surgical intervention is associated with recurrence\nrates of approximately 20%,  while hormonal\nmanipulation frequently induces adverse effects due to steroid hormone\nimbalance. \n , \n  Given the significant impact of endometriosis\non the quality of life and the limitations of existing treatments,\nwhich primarily target symptom control rather than the underlying\npathophysiology, there is a critical need for the development of new\ntherapeutic agents to improve disease management.\nNatural products\nconstitute an invaluable source of therapeutic\nagents, owing to their remarkable structural diversity. According\nto Newman and Cragg (2020), approximately 48% of all new drugs approved\nby the FDA between 1981 and 2019 were derived from natural sources.  Among these, phenolic compounds represent a highly\ndiverse group of plant metabolites recognized for their broad biological\npotential, particularly their antioxidant, anti-inflammatory, and\nanticancer activities. \n −\nFlavonoids, in particular, are capable of scavenging free\nradicals\nand reducing oxidative stress, while modulating the activity of several\nenzymes and signaling pathways, thereby reinforcing their protective\neffects on human health. \n − \n \n \n \n \n \n \n \n  Due to their versatile reactivity, chalcones have emerged as privileged\nchemical scaffolds, serving as key intermediates in the design of\na wide range of biologically active derivatives exhibiting anti-inflammatory,\nantihistaminic, antioxidant, antiobesity, antiparasitic, and other\npharmacological properties. \n , −\nDespite these advances, experimental studies addressing endometriosis\nat the cellular level remain limited, while a substantial body of\nrecent research has focused on patient-based and clinical investigations. \n , \n  Most in vitro investigations have focused on inflammatory signaling,\noxidative stress, or hormonal responsiveness, frequently relying on\nsingle-cell-line systems or reporting general cytotoxic effects without\ndirect comparison between endometriotic and eutopic endometrial cells.\nImportantly, a small number of established cellular models continue\nto underpin most experimental work in this area, and comparative screening\napproaches have changed little over time. As a result, systematic\nevaluations of differential cytotoxic responses and selectivity in\nendometriosis-relevant cellular models are still scarce. \n ,\nIn this context, the present study aimed to evaluate the cytotoxic\nand selective properties of a series of different flavonoid derivatives\n( 1 – 37 ) as an initial screening step,\nusing models of normal endometrial (Ishikawa) and endometriotic (12Z)\ncells. \n , \n  The investigation focused on correlating\nkey physicochemical descriptors, such as lipophilicity (log  P ), topological polar surface area (TPSA), and hydrogen-bonding\ncapacity, with the observed biological responses to identify patterns\nassociated with cytoprotective versus cytotoxic selectivity.\n\nA set of flavonoid\nderivatives ( 1 – 37 ,  Figure  \n ) had their\ncytotoxicity evaluated for the Ishikawa cell line, commonly used to\nsimulate the normal endometrium, by the MTT assay.\nStructures of flavonoid\nderivatives  1 – 37  evaluated for cytotoxicity\nin Ishikawa and 12Z cell lines.\nThe viability assay conducted in Ishikawa cells\n( Figure  \n ) revealed\na broad spectrum\nof biological responses across the compound library. A subset of molecules\nexhibited cytoprotective activity, with viabilities exceeding 100%,\nincluding  1  (155%),  2  (150%),  6  (102%),  7  (149%),  8  (126%),  17  (127%),  28  (114%), and  30  (114%). These\nelevated viability values may reflect enhanced epithelial cell stratification\nor increased mitochondrial activity, as detected by the MTT assay\nemployed. Considering chemical aspects, several compounds correspond\nto polyhydroxylated or methoxylated flavonoids (e.g.,  1 ,  2 , and  7 ), whereas others possess fewer\npolar substituents ( 8 ) or halogen groups ( 28 ). This result suggests that cytoprotective effects are not solely\ndetermined by the number of hydroxyl or methoxyl groups but rather\nby the nature, position, and overall electronic interplay of substituents\nwithin the scaffold. A second group of molecules displayed nontoxic\nbehavior, maintaining viabilities between 75% and 100%, including  5  (83%),  13  (87%),  14  (82%),  15  (78%),  23  (84%), and  26  (81%).\nThese derivatives preserved near-baseline cell viability without a\nconsistent substitution pattern, indicating that no single substituent\nclass could be identified as a dominant determinant of cytoprotective\nbehavior within the present data set.\nCell viability of Ishikawa cells after\nexposure to compounds  1 – 37  at 200\nμM for 24 h. Data are\nexpressed as means ± SD relative to the negative control (untreated\ncells, 100%). Viability was determined by the MTT assay.  N  = 3 independent experiments performed in triplicate. Statistical\nsignificance versus the negative control was assessed by one-way ANOVA\nwith Bonferroni’s post-test, with significance represented\nas  p  < 0.05 (*) and  p  < 0.001\n(***).\nConversely, several compounds exhibited cytotoxicity\ntoward Ishikawa\ncells such as  3  (65%),  4  (63%),  9  (58%),  10  (53%),  11  (32%), and  29  (12%). This group included compound  18 , the only unsubstituted\nchalcone in the series, as well as derivatives containing electron-withdrawing\nsubstituents such as −NO 2  ( 16 ,  22 ,  34 , and  36 ) or −Cl groups\n( 25  and  33 ). Other compounds, including  29 ,  31 ,  32 , and  37 ,\nalso showed high cytotoxicity, possibly due to the absence of polar\ndonor substituents. Having established an initial safety profile in\nIshikawa cells, subsequent assays were performed to evaluate the activity\nin 12Z cells.\nIn 12Z cells, several compounds exhibited pronounced\ninhibitory\neffects ( Figure  \n ),\nwith the greatest reductions in viability observed for  24  (49%),  15  (51%),  28  (53%),  23  (59%), and  14  (60%). These derivatives, which reduced\nthe viability to approximately 60% or below, stand out as promising\nlead candidates for targeting endometriotic cells. A second group\ndemonstrated moderate inhibition, maintaining viability between 70\nand 90%, such as  30  (70%),  17  (74%),  5  (82%),  6  (86%), and  8  (88%). Although\nless potent, these compounds remain noteworthy when considering their\nfixed-dose selectivity, defined here by comparing 12Z and Ishikawa\ncell viability under identical exposure conditions.\nCell viability of 12Z\ncells after exposure to compounds  1 ,  2 ,  5 – 8 ,  12 – 15 ,  17 ,  23 ,  24 ,  26 ,  28 , and  30  at\n200 μM for 24 h. Data are expressed as means ± SD relative\nto untreated cells (100%). Viability was determined by the MTT assay.  N  = 3 independent experiments performed in triplicate. Statistical\nsignificance versus the control was assessed by one-way ANOVA with\nBonferroni’s post-test with significance represented as  p  < 0.05 (*),  p  < 0.01 (**),  p  < 0.001 (***), and  p  < 0.0001 (****).\nIn contrast, certain compounds produced cytoprotective\nor neutral\neffects in 12Z cells, with viability values exceeding 100%, indicating\nundesirable selectivity in this model, as they may promote the survival\nof diseased cells. Overall, the 12Z data identify compounds  14 ,  15 ,  23 ,  24 , and  28  as the most effective inhibitors of endometriotic cell\nviability, while also highlighting a subset of derivatives that are\ninactive or potentially counterproductive due to their prosurvival\neffects.\nSelectivity between the two cell lines was evaluated\nby directly\ncomparing the viability at a fixed concentration (200 μM, 24\nh). The underlying rationale was that ideal candidates should effectively\nreduce 12Z viability while sparing Ishikawa cells, thereby combining\npotency with safety. Statistical analysis revealed that compounds  1 ,  2 ,  14 ,  17 ,  23 ,  24 ,  28 , and  30  exhibited\nsignificant differences between the two cell lines. Among these,  24  (49% in 12Z vs 109% in Ishikawa,  p  <\n0.001) and  28  (53% vs 115%,  p  < 0.0001)\ndemonstrated the highest selectivity, markedly reducing 12Z viability\nwhile preserving or even enhancing Ishikawa viability, representing\ncytotoxic selectivity (preferential 12Z cell death).\nCompounds  14 ,  17 ,  23 , and  30  also\nshowed statistically supported selectivity, albeit\nwith smaller margins, maintaining Ishikawa viability above 80% while\nreducing 12Z viability into the 50–70% range. Conversely, compounds  1  and  2 , although displaying statistically significant\ndifferences ( p  < 0.0001 and  p  = 0.0101, respectively), retained high 12Z viability (>90%),\nindicating\ncytoprotective selectivity (preferential protection of Ishikawa rather\nthan elimination of 12Z). A secondary group, including compounds  6 – 8  and  15 , exhibited a trend\ntoward selectivity, showing consistent but statistically nonsignificant\ndifferences between the two cell lines.\nImportantly, these results\nhighlight two biologically distinct\npatterns of selectivity at this fixed concentration: (i) preferential\nimpairment of 12Z viability, the desired therapeutic profile, as observed\nfor compounds  24  and  28 ; and (ii) preferential\nprotection of Ishikawa cells, a context-dependent and less therapeutically\nfavorable pattern, as seen for compounds  1  and  7 . Collectively, these findings identify compounds  24  and  28  as the most promising selective agents, with  14 ,  17 ,  23 , and  30  also\nemerging as statistically validated hits.\nAlthough evaluated\nat a single concentration, this analysis provides\na robust framework for assessing therapeutic selectivity: compounds\nthat reduce 12Z viability while maintaining Ishikawa viability are\nconsidered to be the most promising candidates.  Table  \n  summarizes the viability data for both cell\nlines alongside calculated physicochemical descriptors (log  P , TPSA, HBD, and HBA), providing the data set underlying\nthe selectivity assessment ( Figure  \n ) and subsequent structure–activity relationship\nanalysis.\nComparison of cell viability between Ishikawa, indicated as  1 , and 12Z, indicated as  2 , cells after exposure\nto selected compounds ( 1 ,  2 ,  5 – 8 ,  12 – 15 ,  17 ,  23 ,  24 ,  26 ,  28 , and  30 ) at 200 μM for 24 h. Data are\nexpressed as means ± SD relative to untreated cells (100%). Statistical\nanalysis was performed using two-way ANOVA with Bonferroni’s\npost-test with  p  < 0.05 (*),  p  < 0.01 (**),  p  < 0.001 (***), and  p  < 0.0001 (****).\nΔviability represents the\ndifference between Ishikawa and 12Z viability. Physicochemical descriptors\n(log  P , TPSA, HBD, and HBA) were calculated using\nRDKit.  N  = 3 independent experiments performed in\ntriplicate.\nCluster-based analysis was used as a descriptive tool\nto group\ncompounds according to similarity in their physicochemical descriptors\nand observed biological responses ( Figure  \n ). Cluster 1 comprised molecules bearing\noxygenated substituents (−OH and −OMe) and nitro groups\n(−NO 2 ), which displayed intermediate toxicity in\nboth cell lines (compounds  14 ,  15 ,  16 ,  17 ,  28 ,  29 , and  35 ). This cluster reflects a recurring phenotypic response\npattern within the investigated chemical space, rather than a definitive\nstructure–activity relationship, and is associated with limited\nfixed-dose selectivity.\nStructural clustering\nof the compound library. Hierarchical clustering\n(Ward method) separated the compounds into six clusters (C1–C6).\nEach cluster reflects recurring phenotypic response patterns observed\nat a fixed dose: C1–C2 grouped moderately active derivatives,\nC3 concentrated the most cytotoxic compounds (cell line-dependent),\nC4 and C6 contained polyhydroxylated/methoxylated structures associated\nwith cytoprotective responses, and C5 comprised simpler molecules\nwith intermediate viability.\nCluster 2, enriched with −NH 2  and −NMe 2  derivatives (compounds  21 – 27 ), exhibited modest cytotoxicity and tended\ntoward cytoprotective\nselectivity, maintaining high tolerability in both cell lines, typically\nwith an Ishikawa cell viability of ≥80%. In contrast, cluster\n3, composed of nitro- and halogen-substituted chalcones (compounds  18 – 20 ,  30 – 34 ,  36 , and  37 ), included the most cytotoxic\ncompounds toward Ishikawa cells. This cluster displayed cytotoxic-type\nselectivity when the 12Z viability was more strongly reduced than\nIshikawa at the fixed dose.\nClusters 4 (compounds  1 ,  2 ,  3 ,  6 ,  8 ,  9 , and  13 ) and 6 (compounds  7 ,  10 ,  11 , and  12 ), dominated by polyhydroxylated\nand methoxylated\nflavonoids, consistently maintained high viability (often >100%),\ncorresponding to cytoprotective selectivity that is favorable for\nIshikawa cells but undesirable for 12Z. Cluster 5 comprised flavanones\n(compounds  4  and  5 ) featuring a partially\nsaturated benzopyranone core, which displayed intermediate but largely\nnonselective biological profiles.\nOverall, these findings indicate\nthat selectivity within this compound\nset is best described by phenotypic response patterns associated with\nphysicochemical properties, rather than by the scaffold type or mechanistic\nelectronic interpretations.\nAnalysis of lipophilicity (log  P ), calculated\nfrom RDKit descriptors, revealed an empirical association between\nhigher log  P  values and reduced cell viability, particularly\nin 12Z cells. Compounds in the upper quartile (log  P  > 3.5) were enriched among the more cytotoxic series, including\nmembers of clusters 1 and 3 (nitro- and halogen-substituted chalcones),\nwhich were among the compounds exhibiting stronger cytotoxic effects. \n , \n  In contrast, polyhydroxylated and methoxylated clusters (4 and 6)\ndisplayed significantly lower log  P  values (<3),\ncorresponding to reduced cytotoxicity and, in some cases, proliferative\neffects consistent with polyphenol-associated cytoprotection. \n , \n  Within the fixed-dose selectivity framework, higher log  P  values were associated with a cytotoxic selectivity profile,\ncharacterized by preferential reduction of 12Z viability with preservation\nof Ishikawa cells, as exemplified by compounds  28  (log  P  = 3.96; 12Z ≈ 53%; Ishikawa ≈ 115%) and  24  (log  P  = 3.35; 12Z ≈ 49%; Ishikawa\n≈ 109%). Conversely, lower log  P  values combined\nwith multiple hydroxyl or methoxyl substituents (e.g., compound  1 , log  P  = 2.84) were associated with cytoprotective\nbehavior, preserving Ishikawa viability while only modestly affecting\n12Z cells.\nThe topological polar surface area (TPSA) was analyzed\nas an independent\ndescriptor, with higher TPSA values associated with reduced cytotoxicity\nat the fixed dose. \n , \n  Compounds with TPSA > 80 Å 2  (compounds  1 ,  2 ,  6 ,  7 ,  12 , and  13 ), indicative\nof cytoprotective selectivity, are favorable for Ishikawa but undesirable\nfor 12Z. In contrast, TPSA < 40 Å 2  (compounds  28 ,  29 ,  32 ,  33 ,  34 , and  36 ) was associated with greater cytotoxicity\nand, when the Ishikawa viability remained ≥80%, reflected cytotoxic-type\nselectivity. For example, compound  12  (TPSA ≈\n137 Å 2 ) exhibited high polarity and cytoprotective\nbias (Ishikawa ≈ 70%; 12Z ≈ 120%), while compound  28 , with low TPSA and high log  P , displayed\npreferential 12Z inhibition. Thus, under fixed-dose conditions, high\nTPSA favored cytoprotective selectivity, whereas low TPSA, particularly\nwhen accompanied by higher lipophilicity, was more frequently observed\namong compounds showing cytotoxic selectivity.\nHydrogen-bonding\nparameters, expressed as the number of hydrogen-bond\ndonors (HBD) and acceptors (HBA), paralleled TPSA trends and further\nrefined this interpretation. Polyhydroxylated derivatives (compounds  1 ,  2 ,  6 ,  7 ,  12 , and  13 ), characterized by multiple hydrogen-bond donor\nand acceptor functionalities, consistently exhibited reduced cytotoxicity\nin both cell lines, frequently resulting in cytoprotective selectivity. \n , , \n  In contrast, chalcones with fewer\nhydrogen-bonding features (compounds  28 ,  29 ,  32 – 34 , and  36 ) displayed\nincreased cytotoxicity and, when Ishikawa viability remained above\n80%, demonstrated cytotoxic-type selectivity toward 12Z. Compound  12 , which presents multiple hydrogen-bonding functionalities,\nexemplified relative 12Z sparing compared with Ishikawa cells, whereas\ncompound  28  (HBD = 0; HBA = 3) illustrated the opposite\nbehavior, representing two extremes within the observed selectivity\nspectrum.\nTo integrate these relationships, a correlation heatmap\nwas generated\n( Figure  \n ). log  P  correlated negatively with 12Z viability ( r  = −0.73), whereas TPSA, HBD, and HBA showed strong positive\ncorrelations ( r  = 0.77–0.83). Ishikawa viability\nexhibited only weak correlations ( r  ≤ 0.26).\nImportantly, Δviability (Ishikawa – 12Z) demonstrated\nweak, but consistent trends, negative with log  P  ( r  = −0.24) and positive with TPSA/HBD/HBA ( r  = 0.19–0.25), supporting the dual-mode selectivity\nmodel: higher lipophilicity drives cytotoxic selectivity toward 12Z,\nwhile higher polarity and hydrogen-bonding capacity favor cytoprotective\nselectivity toward Ishikawa.\nCorrelation heatmap of physicochemical descriptors\nand biological\noutcomes for compounds  1 ,  2 ,  5 – 8 ,  12 – 15 ,  17 ,  23 ,  24 ,  26 ,  28 , and  30 . Pearson correlation coefficients\n( r ) are represented by a red–blue scale, with\nred indicating negative and blue indicating positive correlations\n(−1 to +1). Lipophilicity (log  P ) showed a\nstrong negative correlation with 12Z viability ( r  = −0.73), whereas polarity-related descriptors (TPSA, HBD,\nand HBA) correlated positively ( r  = 0.77–0.83).\nIn contrast, Ishikawa cell viability exhibited no significant association\nwith the descriptors ( r  ≤ 0.26). Δviability\n(Ishikawa – 12Z) demonstrated weak but consistent correlations,\nbeing negatively associated with log  P  ( r  = −0.24) and positively associated with TPSA, HBD, and HBA\n( r  = 0.19–0.25).\nFinally, the molecular weight and ring count showed\nminimal variation\n(approximately 250–300 Da and two to three aromatic rings,\nrespectively) and showed no meaningful correlation with either activity\nor selectivity. This indicates that within the studied chemical space,\nsubstituent-dependent parameters, including lipophilicity, polarity,\nhydrogen bonding, and electronic effects, are the primary determinants\nof fixed-dose selectivity outcomes. Overall, chalcone represents a\ntunable molecular scaffold in which subtle substituent modifications\ndictate whether selectivity follows a cytotoxic pathway (desired:\nreduced 12Z viability with preserved Ishikawa viability) or a cytoprotective\npathway (undesired for therapy: high viability for both cell lines).\nAcross the series, compounds bearing nitro or chloro substituents\n(−NO 2  and −Cl) and exhibiting higher lipophilicity\nwere more frequently associated with reduced 12Z viability while maintaining\na higher Ishikawa cell viability at the fixed dose. In contrast, compounds\ncontaining hydroxyl, methoxy, or amino substituents (−OH, −OMe,\nand −NR 2 ), together with increased polarity-related\ndescriptors such as TPSA and hydrogen-bonding capacity, tended to\ndisplay lower cytotoxicity and, in some cases, cytoprotective or proliferative\neffects. This fixed-dose evaluation, based on the comparative response\nof endometriotic (12Z) and eutopic endometrial (Ishikawa) cells, provides\na consistent operational definition of selectivity throughout the\nstudy and explains why compounds  24  and  28  emerge as the most selective candidates, with  14 ,  17 ,  23 , and  30  identified as additional\nstatistically supported hits.\n\nNuclear magnetic resonance\n(NMR) spectra were obtained using a Varian INOVA spectrometer, operating\nat 500 and 125 MHz for  1 H and  13 C nuclei, respectively,\nwith CDCl 3  or DMSO- d \n 6  or CD 3 OD (Sigma-Aldrich) as the solvent and internal standards.\nFlavonoids  1 – 4 ,  7 – 8 , and  10 – 11  were obtained commercially (Sigma-Aldrich, purity ≥95%)\nand used without further purification. Flavonoids  5 ,  9 , and  12  were isolated from the DCM phase of\nMeOH extract from the leaves of  Baccharis lateralis , while  6  and  13  were isolated from MeOH\nextract from the leaves of  Baccharis sphenophylla , as previously reported. \n , \n  Compound identities\nwere confirmed by comparison of their NMR data with reported literature\nvalues.\nChalcones  14 – 37  were prepared by Claisen–Schmidt condensation as previously\ndescribed. \n , \n  Compound identities were confirmed\nby comparison of their NMR data with reported literature values.\nIshikawa and 12Z cells\n(Sigma-Aldrich) were cultured in 75 cm 2  flasks at 37 °C\nand 5% CO 2 . Ishikawa cells used MEM with 5% FBS, and 12Z\ncells used high-glucose DMEM with 10% FBS. Both media were supplemented\nwith penicillin (100 U/mL), streptomycin (10 U/mL), nonessential amino\nacids, and sodium pyruvate.\nThe MTT assay\nwas employed to determine the cell viability. Ishikawa and 12Z cells\nwere plated in 96-well plates at 4 × 10 4  cells/cm 2  and incubated for 24 h. Subsequently, cells were treated\nwith compounds  1 – 37  at a concentration\nof 200 μM for 24 h. Stock solutions of all compounds were prepared\nin DMSO and diluted in culture medium to the desired concentration,\nresulting in a final DMSO concentration of 1% (v/v) in all treatments;\ncontrol wells contained the same DMSO concentration. After that, 30\nμL of MTT solution (5 mg/mL) was added to each well, and the\nplates were kept in the dark at 37 °C for 2 h. Results were expressed\nas cell viability % in comparison to untreated cells. All experiments\nwere conducted in triplicate as independent assays. Analysis of variance\n(ANOVA) with Bonferroni’s test was used to evaluate the differences\nbetween cell groups (negative control versus treatment) for MTT experiments\nwith a level of significance set at  p  < 0.05.\nThe chemical structures of compounds  1 – 37  were represented using SMILES notation and processed with\nthe RDKit cheminformatics package (version 2024.09.6).  From these structures, key physicochemical descriptors\nwere calculated, including lipophilicity (log  P ,\nestimated by Crippen’s atom-additive MolLog P model and reported\nas dimensionless), topological polar surface area (TPSA, in Å 2 ), number of hydrogen-bond donors (HBD) and acceptors (HBA),\nmolecular weight (MW, Da), and total ring count (aromatic and aliphatic). \n , \n  All descriptor values were exported and are compiled in the  Supporting Information, Table S1  to support structure–activity\nrelationship (SAR) analysis.\nThe selectivity index was calculated\nas the difference in cell viability between Ishikawa and 12Z cells\n(Δviability = Ishikawa – 12Z), with positive values indicating\nselective toxicity toward 12Z cells. Correlations between physicochemical\ndescriptors and biological activity were evaluated using Pearson’s\ncorrelation coefficient ( r ), and the corresponding\ncoefficients of determination ( R \n 2 ) were\ncomputed using GraphPad Prism 10 and were represented as a heatmap. \n ,\nHierarchical clustering was performed in Python (scipy.cluster.hierarchy,\nv1.14.1) using Ward’s linkage and Euclidean distance to examine\nsimilarity relationships among the compounds based on the selected\nphysicochemical descriptors.  The resulting\ndendrogram was used to define six clusters, which were subsequently\nemployed as a descriptive framework to organize and discuss the observed\nbiological response patterns.  Structural\nfeatures, including common substituents, are described within each\ncluster solely to aid qualitative interpretation without being used\nas criteria for cluster definition or boundary assignment.\n\nIn this study, we successfully\nevaluated a library of 37 flavonoid\nderivatives, primarily chalcones, for their potential as selective\ntherapeutic agents against endometriosis. By employing a comparative\nscreening approach using 12Z endometriotic cells and Ishikawa normal\nendometrial cells, differential biological response patterns were\nobserved across the compound series. These findings indicate that\nvariations in physicochemical properties are associated with shifts\nbetween cytotoxic and cytoprotective phenotypes without supporting\nthe establishment of a distinct structure–activity relationship.\nDerivatives featuring multiple hydroxyl or methoxy groups, high polarity\n(TPSA > 80 Å 2 ), and extensive hydrogen-bonding\ncapacity\nconsistently exhibited cytoprotective effects, particularly in normal\nendometrial cells. Conversely, several chalcones bearing nitro or\nchloro substituents and exhibiting higher lipophilicity were among\nthe compounds that showed increased cytotoxicity at a fixed dose.\nAmong the evaluated compounds, compounds  24  and  28  emerged as the most selective within the series. At the\nfixed dose, both compounds reduced the viability of endometriotic\n12Z cells to approximately 50%, while preserving or, in some cases,\nincreasing the viability of Ishikawa cells. This differential response\ncharacterizes a cytotoxic selectivity profile, in which activity is\npreferentially directed toward endometriotic cells rather than reflecting\nnonspecific cytotoxicity. Such a profile is particularly relevant\nin the context of endometriosis, where selective targeting of ectopic\nendometrial tissues remains a major therapeutic challenge.\nCollectively,\nthis work highlights chalcones as versatile scaffolds\nfor the exploration of bioactive compounds with selective effects\nin endometriosis-related cellular models. The insights gained from\nthe correlation between physicochemical descriptors (log  P  and TPSA) and biological outcomes provide a useful framework to\nguide future exploratory studies aimed at identifying more potent\nand selective compounds, paving the way for novel, nonhormonal therapeutic\nstrategies for endometriosis management.","source_license":"CC-BY-4.0","license_restricted":false}