血藥濃度-時間曲線圖（藥時曲線）是藥物被給予機體后，以時間為橫坐標，藥物（及其代謝物）在血液中的濃度為縱坐標所繪制的血藥濃度隨時間變化的曲線。
藥時曲線能反映體內藥物濃度的經時變化過程。為設計采血點、給藥方案提供依據。因此繪制藥時曲線是藥代動力學分析中的一個重要部分。
根據CDE頒布的《藥物臨床試驗數據遞交指導原則（試行）》中指出申報資料中的數據集通常采用SAS數據傳輸格式(SAS Transport Format，簡稱XPT)，由此可見SAS軟件在藥品臨床試驗數據分析中是必不可缺。
其中SAS Graph Template Language (GTL)是繪制復雜圖形的強大工具。本文將使用SAS編程呈現幾種不同的藥時曲線樣式，探索如何繪制更協調更直觀的圖形，供大家共同學習討論！
基于GTL的圖形由控制圖形格式和外觀，并指定變量角色的模板定義生成。圖形是通過關聯模板與數據源來呈現。
1. 使用template過程定義STATGRAPH模板
2. 使用圖形模板語言來指定圖形的參數
proc template;
define statgraph tmp1;
/*利用dynamic將template中的變量、標簽等內容開放出來，便于修改dynamic的值來控制輸出，重復使用*/
dynamic XAXISLABEL YVAR XVAR GRPVAR LENGEDTITLE TICKMIN TICKMAX;
begingraph;
[自定義的圖形模板]
endgraph;
end;
run;
3. 通過使用SGRENDER過程將數據與模板關聯
proc sgrender data=[數據源] template=[GTL模板];
/*不同表僅修改dynamic即可*/
dynamic XAXISLABEL='Time(h)' YVAR='Mean' XVAR='ATPTN' LENGEDTITLE='劑量組：'
GRPVAR='ARMCD' TICKMIN='0.5' TICKMAX='13';
run;
GTL最強大的特性之一是圍繞著分層語句塊（即：Layout布局）構建的語法，常用布局：
1. OVERLAY : 用于在單個單元格中顯示二維繪圖的通用布局
2. GRIDDED : 網格圖形，所有單元格獨立
3. LATTICE : 升級版多單元格布局。坐標軸可以跨列或行共享，并且位于網格外部
proc template;
define statgraph Simple_Layout;
begingraph;
entrytitle 'Simple Layout';
layout lattice / columns=2 columngutter=5;
layout overlay / walldisplay=(outline) xaxisopts=(display=none) yaxisopts=(display=none);
scatterplot x=weight y=systolic / markerattrs=(size=0);
entry halign=center textattrs=(size=20) "1" / valign=center;
endlayout;
layout overlay / walldisplay=(outline) xaxisopts=(display=none) yaxisopts=(display=none);
scatterplot x=weight y=diastolic / markerattrs=(size=0);
entry halign=center textattrs=(size=20) "2" / valign=center;
endlayout;
endlayout;
endgraph;
end;
run;
proc sgrender data=sashelp.heart template=Simple_Layout;
run;
proc template;
define statgraph Complex_Layout;
begingraph;
layout lattice / columns=2 columnweights=(.6 .4) columngutter=5;
layout lattice / rows=4 rowweights=(.25 .2 .2 .35) rowgutter=5;
layout lattice / columns=2 rowweights=(.6 .4) columngutter=5;
layout overlay / walldisplay=(outline) xaxisopts=(display=none) yaxisopts=(display=none);
scatterplot x=height y=weight / markerattrs=(size=0);
entry halign=center textattrs=(size=20) "1" / valign=center;
endlayout;
layout overlay / walldisplay=(outline) xaxisopts=(display=none) yaxisopts=(display=none);
scatterplot x=height y=weight / markerattrs=(size=0);
entry halign=center textattrs=(size=20) "2" / valign=center;
endlayout;
endlayout;
layout overlay / walldisplay=(outline) xaxisopts=(display=none) yaxisopts=(display=none);
scatterplot x=height y=weight / markerattrs=(size=0);
entry halign=center textattrs=(size=20) "3" / valign=center;
endlayout;
layout overlay / walldisplay=(outline) xaxisopts=(display=none) yaxisopts=(display=none);
scatterplot x=height y=weight / markerattrs=(size=0);
entry halign=center textattrs=(size=20) "4" / valign=center;
endlayout;
layout lattice / columns=2 rowweights=(.6 .4) columngutter=5;
layout overlay / walldisplay=(outline) xaxisopts=(display=none) yaxisopts=(display=none);
scatterplot x=height y=weight / markerattrs=(size=0);
entry halign=center textattrs=(size=20) "5" / valign=center;
endlayout;
layout overlay / walldisplay=(outline) xaxisopts=(display=none) yaxisopts=(display=none);
scatterplot x=height y=weight / markerattrs=(size=0);
entry halign=center textattrs=(size=20) "6" / valign=center;
endlayout;
endlayout;
endlayout;
layout overlay / walldisplay=(outline) xaxisopts=(display=none) yaxisopts=(display=none);
scatterplot x=height y=weight / markerattrs=(size=0);
entry halign=center textattrs=(size=20) "7" / valign=center;
endlayout;
endlayout;
endgraph;
end;
run;
ods graphics / reset width=5in height=5in imagename='Complex_Layout';
proc sgrender data=sashelp.class template=Complex_Layout;
run;
準備一份不同采血點的血藥濃度Mean±SD的模擬數據：
data test;
input ARMCD $10. ATPTN Mean Meanlow Meanhig;
datalines;
劑量組1 1 212.5 206.1 218.9
劑量組1 2 141.9 66.8 217.0
劑量組1 3 118.7 50.4 187.0
劑量組1 4 87.4 41.3 133.5
劑量組1 6 58.4 26.9 89.8
劑量組1 8 40.7 15.7 65.6
劑量組1 12 19.9 7.9 31.9
劑量組2 1 87.7 23.5 151.8
劑量組2 2 135.3 135.3 290.5
劑量組2 3 126.4 126.4 282.8
劑量組2 4 113.3 113.3 250.1
劑量組2 6 81.9 81.9 175.4
劑量組2 8 72.0 4.1 139.9
劑量組2 12 45.5 16.5 74.4
劑量組3 1 355.5 83.3 627.7
劑量組3 2 290.5 107.4 473.6
劑量組3 3 303.0 206.8 399.2
劑量組3 4 217.0 163.3 270.7
劑量組3 6 166.0 110.8 221.2
劑量組3 8 129.5 92.0 167.0
劑量組3 12 84.4 53.7 115.0
;
run;
藥時曲線的血藥濃度所對應的坐標軸通常采用線性和半對數兩種不同方式呈現。
1. 使用lattice作為最外層布局
2. 內嵌兩個overlay作為基礎單元格布局，分別容納線性藥時曲線和半對數藥時曲線
proc template;
define statgraph tmp1;
dynamic XAXISLABEL YVAR XVAR GRPVAR LENGEDTITLE TICKMIN TICKMAX;
begingraph;
layout lattice / columns=2 rows=1;
column2headers;
entry "線性藥時曲線";
entry "半對數藥時曲線";
endcolumn2headers;
layout overlay/
xaxisopts=(label=XAXISLABEL offsetmin=0 offsetmax=0
linearopts=(tickvaluelist=(0 1 2 3 4 6 7 8 12) viewmin=TICKMIN viewmax=TICKMAX TICKVALUEFITPOLICY=ROTATE)
tickvalueattrs=(size=7pt) labelattrs=(size=7pt))
yaxisopts=(label="Concentration (ng/mL)") walldisplay=none ;
seriesplot y=YVAR x=XVAR / group=GRPVAR name="stocks1";
scatterplot y=YVAR x=XVAR / group=GRPVAR name="stocks2"
YERRORLOWER=meanlow YERRORUPPER=meanhig markerattrs=(symbol=squarefilled size=8) ;
mergedlegend "stocks1" "stocks2" /title=LENGEDTITLE border=true across=4 DISPLAYCLIPPED=TRUE;
endlayout;
layout overlay/
xaxisopts=(label=XAXISLABEL offsetmin=0 offsetmax=0
linearopts=(tickvaluelist=(0 1 2 3 4 6 7 8 12) viewmin=TICKMIN viewmax=TICKMAX TICKVALUEFITPOLICY=ROTATE)
tickvalueattrs=(size=7pt) labelattrs=(size=7pt))
yaxisopts=(label="Concentration (ng/mL)" TYPE=LOG LOGOPTS=(base=10)) walldisplay=none ;
seriesplot y=YVAR x=XVAR / group=GRPVAR name="stocks3";
scatterplot y=YVAR x=XVAR / group=GRPVAR name="stocks4"
YERRORLOWER=meanlow2 YERRORUPPER=meanhig markerattrs=(symbol=squarefilled size=8) ;
mergedlegend "stocks3" "stocks4"/title=LENGEDTITLE /*border=true*/ across=4 DISPLAYCLIPPED=TRUE;
endlayout;
endlayout;
endgraph;
end;
run;
proc sgrender data=test template=tmp1;
dynamic XAXISLABEL='Time(h)' YVAR='Mean' XVAR='ATPTN' LENGEDTITLE='劑量組：' GRPVAR='ARMCD' TICKMIN='0.5' TICKMAX='13';
run;
將線性藥時曲線和半對數曲線直接橫向排列起來是最簡單省力的方法，同時也有一些缺點，如：
? 圖例重復
? 圖形中大面積空白，缺少層次
藥時曲線的消除相部分有大量空白，嘗試將半對數藥時曲線縮小填充到空白部分，提高圖形層次感。
1. 使用overlay作為最外層布局，容納線性藥時曲線
2. 內嵌gridded作為獨立的單元格布局，嵌入半對數藥時曲線
解決了
? 圖例重復
? 圖形合并
但不同分組標記形狀一致，僅以顏色來區分的效果較差。
proc template;
define statgraph tmp2;
dynamic XAXISLABEL YVAR XVAR GRPVAR LENGEDTITLE TICKMIN TICKMAX;
begingraph;
layout overlay/
xaxisopts=(label=XAXISLABEL offsetmin=0 offsetmax=0
linearopts=(tickvaluelist=(1 2 3 4 6 7 8 12) viewmin=TICKMIN viewmax=TICKMAX TICKVALUEFITPOLICY=ROTATE)
tickvalueattrs=(size=7pt) labelattrs=(size=7pt))
yaxisopts=(label="Concentration (ng/mL)" TYPE=LINEAR) ;
seriesplot y=YVAR x=XVAR / group=GRPVAR name="scatter";
scatterplot y=YVAR x=XVAR / group=GRPVAR name="scatter2"
YERRORLOWER=meanlow YERRORUPPER=meanhig ;
layout gridded / width=300px height=200px halign=right valign=top;
layout overlay/
xaxisopts=(label=XAXISLABEL offsetmin=0 offsetmax=0
linearopts=(tickvaluelist=(1 2 3 4 6 7 8 12) viewmin=TICKMIN viewmax=TICKMAX TICKVALUEFITPOLICY=ROTATE)
tickvalueattrs=(size=7pt) labelattrs=(size=7pt))
yaxisopts=(label="Concentration (ng/mL)" TYPE=LOG
LOGOPTS=(base=10 tickvaluelist=(1 10 100 1000) viewmin=0 viewmax=1000));
seriesplot y=YVAR x=XVAR / group=GRPVAR;
scatterplot y=YVAR x=XVAR / group=GRPVAR
YERRORLOWER=meanlow YERRORUPPER=meanhig ;
endlayout;
endlayout;
mergedlegend "scatter" "scatter2"/title=LENGEDTITLE across=4 DISPLAYCLIPPED=TRUE;
endlayout;
endgraph;
end;
run;
proc sgrender data=test template=tmp2 ;
dynamic XAXISLABEL='Time(h)' YVAR='Mean' XVAR='ATPTN' LENGEDTITLE='劑量組：' GRPVAR='ARMCD' TICKMIN='0.5' TICKMAX='13';
run;
1. 添加discreteattrmap塊，設置不同分組的顯示屬性，讓區分更具辨識度
discreteattrmap name="symbols" / ignorecase=true;
value "XXX" / markerattrs=() lineattrs=() TEXTATTRS=();
enddiscreteattrmap;
discreteattrvar attrvar=groupmarkers var=[分組變量] attrmap="symbols";
2. 在半對數藥時曲線中使用entry語句添加類似水印效果的文字
entry "LOG SCALE" /valign=bottom TEXTATTRS=(COLOR=CXDCDCDC SIZE=15 WEIGHT=BOLD);
proc template;
define statgraph tmp3;
dynamic XAXISLABEL YVAR XVAR GRPVAR LENGEDTITLE TICKMIN TICKMAX;
begingraph;
discreteattrmap name="symbols" / ignorecase=true;
value "劑量組1" / markerattrs=(symbol=triangleFilled color=blue size=9 transparency = 0.6) lineattrs=(color=blue THICKNESS=0.5 PATTERN=4) TEXTATTRS=(size = 11pt family='Times New Roman');
value "劑量組2" / markerattrs=(symbol=circleFilled color=green size=9 transparency = 0.6) lineattrs=(color=green THICKNESS=0.5 PATTERN=4) ;
value "劑量組3" / markerattrs=(symbol=squareFilled color=firebrick size=9 transparency = 0.6) lineattrs=(color=firebrick THICKNESS=0.5 PATTERN=4);
enddiscreteattrmap;
discreteattrvar attrvar=groupmarkers var=GRPVAR attrmap="symbols";
layout overlay/
xaxisopts=(label=XAXISLABEL offsetmin=0 offsetmax=0
linearopts=(tickvaluelist=(1 2 3 4 6 7 8 12) viewmin=TICKMIN viewmax=TICKMAX TICKVALUEFITPOLICY=ROTATE)
tickvalueattrs=(size=7pt) labelattrs=(size=7pt))
yaxisopts=(label="Concentration (ng/mL)" TYPE=LINEAR) ;
seriesplot y=YVAR x=XVAR / group=groupmarkers name="scatter";
scatterplot y=YVAR x=XVAR / group=groupmarkers name="scatter2"
YERRORLOWER=meanlow YERRORUPPER=meanhig ERRORBARATTRS=(THICKNESS=0.5 PATTERN=4);
layout gridded / width=300px height=200px halign=right valign=top;
layout overlay/
xaxisopts=(label=XAXISLABEL offsetmin=0 offsetmax=0
linearopts=(tickvaluelist=(1 2 3 4 6 7 8 12) viewmin=TICKMIN viewmax=TICKMAX TICKVALUEFITPOLICY=ROTATE)
tickvalueattrs=(size=7pt) labelattrs=(size=7pt))
yaxisopts=(label="Concentration (ng/mL)" TYPE=LOG
LOGOPTS=(base=10 tickvaluelist=(1 10 100 1000) viewmin=0 viewmax=1000));
entry "LOG SCALE" /valign=bottom TEXTATTRS=(COLOR=CXDCDCDC SIZE=15 WEIGHT=BOLD);
seriesplot y=YVAR x=XVAR / group=groupmarkers;
scatterplot y=YVAR x=XVAR / group=groupmarkers
YERRORLOWER=meanlow YERRORUPPER=meanhig ERRORBARATTRS=(THICKNESS=0.5 PATTERN=4);
endlayout;
endlayout;
mergedlegend "scatter" "scatter2"/title=LENGEDTITLE across=4 DISPLAYCLIPPED=TRUE;
endlayout;
endgraph;
end;
run;
proc sgrender data=test template=tmp3 ;
dynamic XAXISLABEL='Time(h)' YVAR='Mean' XVAR='ATPTN' LENGEDTITLE='劑量組：' GRPVAR='ARMCD' TICKMIN='0.5' TICKMAX='13';
run;
當前期采血點較為密集導致不同分組的結果不易辨別時，可以將部分時間截取出來放在曲線右上方來增加區分度。
/*===========數據準備=========*/
data test2;
input ARMCD $10. ATPTN Mean Meanlow Meanhig;
datalines;
劑量組1 1 287.1 41.24 532.96
劑量組1 2 343.1 153.44 532.76
劑量組1 3 320.4 175.59 465.21
劑量組1 4 295.9 156.72 435.08
劑量組1 5 268.3 140.22 396.38
劑量組1 6 252.7 136.16 369.24
劑量組1 8 212 114.716 309.284
劑量組1 10 182.1 93.099 271.101
劑量組1 24 84.3 36.133 132.467
劑量組1 48 22.18 8.709 35.651
劑量組1 72 6.207 0.332 12.082
劑量組1 96 0.589 0.589 2.4516
劑量組2 1 388.3 166.36 610.24
劑量組2 2 655.6 390.45 920.75
劑量組2 3 647.4 350.52 944.28
劑量組2 4 608 289.94 926.06
劑量組2 5 548.1 246.31 849.89
劑量組2 6 502.7 222.03 783.37
劑量組2 8 407.7 161.17 654.23
劑量組2 10 346.7 152.92 540.48
劑量組2 24 172.9 76.032 269.768
劑量組2 48 47.56 13.489 81.631
劑量組2 72 13.48 1.219 25.741
劑量組2 96 2.37 2.37 6.5262
劑量組3 1 628.6 355.7 901.5
劑量組3 2 1047 539.45 1554.55
劑量組3 3 1060 585.8 1534.2
劑量組3 4 1008 536.09 1479.91
劑量組3 5 961.7 464.03 1459.37
劑量組3 6 891.4 409.51 1373.29
劑量組3 8 780.2 303.3 1257.1
劑量組3 10 643.9 240.93 1046.87
劑量組3 24 246.6 121.04 372.16
劑量組3 48 60.28 22.872 97.688
劑量組3 72 16.89 3.892 29.888
劑量組3 96 4.387 4.387 9.7029
劑量組4 1 457.3 272.71 641.89
劑量組4 2 924.5 735.7 1113.3
劑量組4 3 1283 818.99 1747.01
劑量組4 4 1550 329.8 2770.2
劑量組4 5 1435 331.3 2538.7
劑量組4 6 1353 365.57 2340.43
劑量組4 8 1101 333.82 1868.18
劑量組4 10 939.3 225.16 1653.44
劑量組4 24 404.8 75.29 734.31
劑量組4 48 103 25.219 180.781
劑量組4 72 24.52 9.958 39.082
劑量組4 96 6.153 1.6598 10.6462
劑量組5 1 1826 370.3 3281.7
劑量組5 2 2700 414.9 4985.1
劑量組5 3 2953 500.4 5405.6
劑量組5 4 2880 733.5 5026.5
劑量組5 5 2593 632.6 4553.4
劑量組5 6 2200 642.6 3757.4
劑量組5 8 1941 488.8 3393.2
劑量組5 10 1896 547.8 3244.2
劑量組5 24 1095 162.49 2027.51
劑量組5 48 421.4 100.4 742.4
劑量組5 72 145 16.36 273.64
劑量組5 96 69.63 9.012 130.248
;
run;
/*調整數據集*/
data test3;
merge test2 test2(where=(ATPTN le 10) rename=(Mean=Mean2 Meanlow=Meanlow2 Meanhig=Meanhig2));
by ARMCD ATPTN;
run;
proc template;
define statgraph tmp4;
dynamic XAXISLABEL YVAR XVAR GRPVAR LENGEDTITLE TICKMIN TICKMAX TICKMAX2;
begingraph;
discreteattrmap name="symbols" / ignorecase=true;
value "劑量組1" / markerattrs=(symbol=triangleFilled color=blue size=9 transparency = 0.6) lineattrs=(color=blue THICKNESS=0.5 PATTERN=4) TEXTATTRS=(size = 11pt family='Times New Roman');
value "劑量組2" / markerattrs=(symbol=circleFilled color=green size=9 transparency = 0.6) lineattrs=(color=green THICKNESS=0.5 PATTERN=4) ;
value "劑量組3" / markerattrs=(symbol=squareFilled color=firebrick size=9 transparency = 0.6) lineattrs=(color=firebrick THICKNESS=0.5 PATTERN=4);
value "劑量組4" / markerattrs=(symbol=HomeDownFilled color=Purple size=9 transparency = 0.6) lineattrs=(color=Purple THICKNESS=0.5 PATTERN=4);
value "劑量組5" / markerattrs=(symbol=StarFilled color=Orange size=9 transparency = 0.6) lineattrs=(color=Orange THICKNESS=0.5 PATTERN=4);
enddiscreteattrmap;
discreteattrvar attrvar=groupmarkers var=GRPVAR attrmap="symbols";
layout overlay/
xaxisopts=(label=XAXISLABEL offsetmin=0 offsetmax=0
linearopts=(tickvaluelist=(1 4 8 10 24 48 72 96) viewmin=TICKMIN viewmax=TICKMAX TICKVALUEFITPOLICY=ROTATE)
tickvalueattrs=(size=7pt) labelattrs=(size=7pt))
yaxisopts=(label="Concentration (ng/mL)" TYPE=LINEAR) ;
seriesplot y=YVAR x=XVAR / group=groupmarkers name="scatter";
scatterplot y=YVAR x=XVAR / group=groupmarkers name="scatter2"
YERRORLOWER=meanlow YERRORUPPER=meanhig ERRORBARATTRS=(THICKNESS=0.5 PATTERN=4);
layout gridded / width=400px height=300px halign=right valign=top;
layout overlay/
xaxisopts=(label=XAXISLABEL offsetmin=0 offsetmax=0
linearopts=(tickvaluelist=(1 2 3 4 6 8 10) viewmin=TICKMIN viewmax=TICKMAX2 TICKVALUEFITPOLICY=ROTATE)
tickvalueattrs=(size=7pt) labelattrs=(size=7pt))
yaxisopts=(label="Concentration (ng/mL)" TYPE=LINEAR) ;
seriesplot y=Mean2 x=XVAR / group=groupmarkers ;
scatterplot y=Mean2 x=XVAR / group=groupmarkers
YERRORLOWER=meanlow2 YERRORUPPER=meanhig2 ERRORBARATTRS=(THICKNESS=0.5 PATTERN=4);
endlayout;
endlayout;
mergedlegend "scatter" "scatter2"/title=LENGEDTITLE across=5 DISPLAYCLIPPED=TRUE;
endlayout;
endgraph;
end;
run;
ods graphics on /width=800px height=400px;
proc sgrender data=test3 template=tmp4 ;
dynamic XAXISLABEL='Time(h)' YVAR='Mean' XVAR='ATPTN' LENGEDTITLE='劑量組：' GRPVAR='ARMCD' TICKMIN='0.5' TICKMAX='100' TICKMAX2='12';
run;
GTL語言具有非常豐富的功能，能夠實現復雜的圖形繪制。僅以本文的藥時曲線為例，其實在美觀性的提升上仍有較大空間。
了解GTL的原理，熟練掌握layout的使用，在實踐中不斷學習、積累和探索，才能繪制更美觀更簡潔的圖形！
熙寧生物|精翰生物已完成臨床藥理服務平臺搭建，正式提供PK、PD、ADA、PK/PD等不同類型臨床項目的統計分析服務。
其中統計編程團隊具有豐富的統計編程經驗，可完成符合CDISC標準的SDTM/ADaM數據集及其相關表格、圖表和列表，與生物分析結合形成一站式服務，歡迎大家后臺留言咨詢交流。
[1] 國家藥品監督管理局，藥物臨床試驗數據遞交指導原則（試行），2020年第16號.
