The next addition to the “strange steroids” series focuses on a couple of steroids possessing anabolic and/or androgenic activity despite the lack of a 17b-hydroxyl function.
The 17b-hydroxyl group is the most fundamental commonality between steroids that attach to and activate the androgen receptor, with all commercialised anabolic steroids either containing, or being metabolized to, a steroid that contains a 17b-OH.
“Without any doubt, the attachment of the steroid to the receptor at the 17-hydroxyl group is one of the most important attachments. The presence of oxygen in the form of a 17b-hydroxyl group is more important than the presence of oxygen at carbon-3 and for practical purposes all useful anabolic agents and highly active androgens possess the 17b-hydroxyl group.” 
The key to the significance of this function is hydrogen bonding, an important feature in ligand-receptor interactions. The Asn705 and Thr877 residues of the androgen receptor form hydrogen bonds with the 17β hydroxyl, locking the steroid into the ligand binding pocket of the receptor.  Modification or elimination of the 17β hydroxyl typically ablates anabolic and androgenic action, but not in every case.
Although prohormones having a ketone function (rather than a hydroxyl) at C17 and steroids esterified at the same position have anabolic and androgenic activity, in both cases they are converted to the ‘active’ hydroxyl form inside the body prior to binding with the androgen receptor (the prohormones by 17β-hydroxysteroid dehydrogenase enzyme, and the esterified steroids by esterase enzymes).
Post Hoc Ergo Log
A team of researchers at the National Research Center in Cairo recently synthesized several unique steroids with appreciable anabolic activity, but which possess nitrogen based groups in lieu of the 17b-hydroxyl group common to AAS. 
The novel N-methyl androgen above was found to have the most favorable Q (anabolic:androgenic) ratio while its N-phenyl analog displayed the highest anabolic activity.
It’s very likely that the steroid is extensively metabolised in the experimental animal prior to interaction with the androgen receptor (as was earlier suggested by the good folks at Ergo-Log).
The European pharmaceutical company Roussel Uclaf synthesized and assayed some 17-deoxysteroids of their own in 1974, and found two of them to still have androgenic properties in the absence of a 17-oxygen. 
According to RU, both compounds retain similar androgenic properties to their 17-oxygenated counterparts. The first is the 17b-deoxy analogue of norethisterone, an oral contraceptive. The second though is described as “a potent androgen” – which is perhaps not entirely unexpected, given that it is the 17b-deoxy analogue of metribolone (“methyl tren”).
There remains the possibility that both of these androgens can be oxygenated in vivo at C-17, thereby effectively acting as “prohormones”, as has been observed with steroids deoxygenated at C3 (for example methylandrostanol – better known as Protobol – can be oxygenated in the body to mestanolone, or methyl DHT ).
Roussel Uclaf didn’t seem to find these 17-deoxysteroids interesting enough to explore further, so the source and extent of their activity has yet to be determined, and they remain a footnote in steroid history.
 Julius A Vida. Androgens and Anabolic Agents. Academic Press. 1969. p. 71-72
 Fragkaki AG, Angelis YS, Koupparis M, Tsantili-Kakoulidou A, Kokotos G, Georgakopoulos C. Structural characteristics of anabolic androgenic steroids contributing to binding to the androgen receptor and to their anabolic and androgenic activities. Applied modifications in the steroidal structure. Steroids. 2009 Feb;74(2):172–197.
 Abd El-Latif NA, Abdulla MM, Amr AE-GE. New steroidal derivatives synthesized using 3beta-hydroxyandrosten-17-one as starting material. Acta Pharm. 2008 Mar;58(1):43–59.
 Active steroids without a 17-hydroxyl function. Synthesis of a new class of progestomimetic agents devoid of androgenic activity. Journal of Steroid Biochemistry vol. 5 issue 4 June, 1974. p. 298
 Wolff ME, Kasuya Y. C-3-Oxygenation of 17-alpha-methyl-5-alpha-androstan-17-beta-ol by rabbit liver homogenate. Journal of Medicinal Chemistry. 1972 Jan;15(1):87–8.