Fatigue-constrained　topology　optimization　(FCTO)　is　a　currently　research　hotspot,　and　its　fatigue　constraints　have　material　property　dependency,　highly　nonlinear,　and　local　features,　which　lead　to　challenges　for　the　algorithm　stability,　computational　efficiency,　and　different　material　application　of　FCTO.　This　research　provides　a　FCTO　method　for　structures　subjected　to　variableamplitude　fatigue　loading,　incorporating　the　potential　orthotropic　behavior　of　materials.　Firstly,　a　fatigue　failure　function　derived　from　the　constitutive　model　of　orthotropic　materials　and　the　polynomial　form　in　the　Tsai-Hill　criterion　is　proposed　to　predict　multiaxial　fatigue　failure　with　a　given　loading　spectrum.　Secondly,　a　FCTO　model　minimizing　structural　weight　is　established　based　on　the　independent　continuous　mapping　(ICM)　method　and　constrained　by　a　filtered,　scaled,　and　aggregated　fatigue　failure　function　to　enhance　stability　and　convergence　speed.　Thirdly,　the　sensitivities　of　objective　and　constraint　in　the　FCTO　model　are　analyzed,　and　the　optimal　model　is　solved　using　convolutional　filters　and　the　globally　convergent　method　of　moving　asymptotes　(GCMMA)　to　generate　manufacturable　design.　Finally,　numerical　examples　demonstrate　the　feasibility　of　the　method　for　2D　and　3D　structures　with　varying　material　properties,　load　spectrums,　and　design　domains.　The　developed　method　aims　to　facilitate　the　creation　of　lightweight　designs　capable　of　withstanding　fatigue　loads　and　to　provide　a　framework　and　references　for　the　advancement　of　integrated　material-structure-performance　designs.